Compare commits

..

3 Commits

Author SHA1 Message Date
19381673a8 More changes for new Linux libs
* Remove unnecessary check for Blosc, this is part of OpenVDB lib now
* Remove WITH_CXX11_ABI, always use new C++11 ABI now
* Replace centos7 by glibc_228 everywhere
2022-12-15 13:18:27 +01:00
140a2b22c1 Temp down-grade GCC 2022-12-15 18:41:41 +11:00
da7bf6ed22 Test building with new 3.5x libraries on Linux 2022-12-15 18:09:40 +11:00
1915 changed files with 36496 additions and 61144 deletions

View File

@@ -61,17 +61,17 @@ ContinuationIndentWidth: 4
# This tries to match Blender's style as much as possible. One
BreakBeforeBraces: Custom
BraceWrapping: {
AfterClass: 'false',
AfterControlStatement: 'false',
AfterEnum : 'false',
AfterFunction : 'true',
AfterNamespace : 'false',
AfterStruct : 'false',
AfterUnion : 'false',
BeforeCatch : 'true',
BeforeElse : 'true',
IndentBraces : 'false',
AfterObjCDeclaration: 'true',
AfterClass: 'false'
AfterControlStatement: 'false'
AfterEnum : 'false'
AfterFunction : 'true'
AfterNamespace : 'false'
AfterStruct : 'false'
AfterUnion : 'false'
BeforeCatch : 'true'
BeforeElse : 'true'
IndentBraces : 'false'
AfterObjCDeclaration: 'true'
}
# For switch statements, indent the cases.

View File

@@ -1,41 +0,0 @@
name: Bug Report
about: File a bug report
labels:
- bug
body:
- type: markdown
attributes:
value: |
### Instructions
First time reporting? See [tips](https://wiki.blender.org/wiki/Process/Bug_Reports) and [walkthrough video](https://www.youtube.com/watch?v=JTD0OJq_rF4).
* Use **Help > Report a Bug** in Blender to fill system information and exact Blender version.
* Test [daily builds](https://builder.blender.org/) to verify if the issue is already fixed.
* Test [previous versions](https://download.blender.org/release/) to find an older working version.
* For feature requests, feedback, questions or build issues, see [communication channels](https://wiki.blender.org/wiki/Communication/Contact#User_Feedback_and_Requests).
* If there are multiple bugs, make multiple bug reports.
- type: textarea
id: body
attributes:
label: "Description"
value: |
**System Information**
Operating system:
Graphics card:
**Blender Version**
Broken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)
Worked: (newest version of Blender that worked as expected)
**Short description of error**
**Exact steps for others to reproduce the error**
Based on the default startup or an attached .blend file (as simple as possible).
- type: markdown
attributes:
value: |
### Help the developers
Bug fixing is important, the developers will handle reports swiftly. For that reason, carefully provide exact steps and a **small and simple .blend file** to reproduce the problem. You do your half of the work, then we do our half!

View File

@@ -1 +0,0 @@
blank_issues_enabled: false

View File

@@ -1,9 +0,0 @@
name: Design
about: Create a design task (for developers only)
labels:
- design
body:
- type: textarea
id: body
attributes:
label: "Description"

View File

@@ -1,9 +0,0 @@
name: To Do
about: Create a to do task (for developers only)
labels:
- todo
body:
- type: textarea
id: body
attributes:
label: "Description"

View File

@@ -1,20 +0,0 @@
name: Pull Request
about: Contribute code to Blender
body:
- type: markdown
attributes:
value: |
### Instructions
Guides to [contributing code](https://wiki.blender.org/index.php/Dev:Doc/Process/Contributing_Code) and effective [code review](https://wiki.blender.org/index.php/Dev:Doc/Tools/Code_Review).
By submitting code here, you agree that the code is (compatible with) GNU GPL v2 or later.
- type: textarea
id: body
attributes:
label: "Description"
value: |
Description of the problem that is addressed in the patch.
Description of the proposed solution and its implementation.

View File

@@ -117,19 +117,15 @@ enable_testing()
# Keep in sync with: https://wiki.blender.org/wiki/Building_Blender
if(CMAKE_COMPILER_IS_GNUCC)
if("${CMAKE_C_COMPILER_VERSION}" VERSION_LESS "11.0.0")
message(FATAL_ERROR "The minimum supported version of GCC is 11.0.0, found ${CMAKE_C_COMPILER_VERSION}")
if("${CMAKE_C_COMPILER_VERSION}" VERSION_LESS "9.3.1")
message(FATAL_ERROR "The minimum supported version of GCC is 9.3.1")
endif()
elseif(CMAKE_C_COMPILER_ID MATCHES "Clang")
if(CMAKE_COMPILER_IS_GNUCC AND ("${CMAKE_C_COMPILER_VERSION}" VERSION_LESS "8.0"))
message(FATAL_ERROR "The minimum supported version of CLANG is 8.0, found ${CMAKE_C_COMPILER_VERSION}")
message(FATAL_ERROR "The minimum supported version of CLANG is 8.0")
endif()
elseif(CMAKE_CXX_COMPILER_ID MATCHES MSVC)
if(MSVC_VERSION VERSION_LESS "1928")
# MSVC_VERSION is an internal version number, it doesn't map to something
# the end user would recognize as a version. Because of this, for MSVC we do
# not show the found number. When using our make.bat the actual VS version
# will be displayed on the console before starting the build, anyway.
message(FATAL_ERROR "The minimum supported version of MSVC is 2019 (16.9.16)")
endif()
endif()
@@ -399,26 +395,6 @@ mark_as_advanced(WITH_SYSTEM_GLOG)
# Freestyle
option(WITH_FREESTYLE "Enable Freestyle (advanced edges rendering)" ON)
# Libraries.
if(UNIX AND NOT APPLE)
# Optionally build without pre-compiled libraries.
# NOTE: this could be supported on all platforms however in practice UNIX is the only platform
# that has good support for detecting installed libraries.
option(WITH_LIBS_PRECOMPILED "\
Detect and link against pre-compiled libraries (typically found under \"../lib/\"). \
Disabling this option will use the system libraries although cached paths \
that point to pre-compiled libraries will be left as-is."
ON
)
mark_as_advanced(WITH_LIBS_PRECOMPILED)
option(WITH_STATIC_LIBS "Try to link with static libraries, as much as possible, to make blender more portable across distributions" OFF)
if(WITH_STATIC_LIBS)
option(WITH_BOOST_ICU "Boost uses ICU library (required for linking with static Boost built with libicu)." OFF)
mark_as_advanced(WITH_BOOST_ICU)
endif()
endif()
# Misc
if(WIN32 OR APPLE)
option(WITH_INPUT_IME "Enable Input Method Editor (IME) for complex Asian character input" ON)
@@ -426,6 +402,11 @@ endif()
option(WITH_INPUT_NDOF "Enable NDOF input devices (SpaceNavigator and friends)" ON)
if(UNIX AND NOT APPLE)
option(WITH_INSTALL_PORTABLE "Install redistributable runtime, otherwise install into CMAKE_INSTALL_PREFIX" ON)
option(WITH_STATIC_LIBS "Try to link with static libraries, as much as possible, to make blender more portable across distributions" OFF)
if(WITH_STATIC_LIBS)
option(WITH_BOOST_ICU "Boost uses ICU library (required for linking with static Boost built with libicu)." OFF)
mark_as_advanced(WITH_BOOST_ICU)
endif()
endif()
option(WITH_PYTHON_INSTALL "Copy system python into the blender install folder" ON)
@@ -520,14 +501,12 @@ endif()
if(NOT APPLE)
option(WITH_CYCLES_DEVICE_ONEAPI "Enable Cycles oneAPI compute support" OFF)
option(WITH_CYCLES_ONEAPI_BINARIES "Enable Ahead-Of-Time compilation for Cycles oneAPI device" OFF)
option(WITH_CYCLES_ONEAPI_HOST_TASK_EXECUTION "Switch target of oneAPI implementation from SYCL devices to Host Task (single thread on CPU). This option is only for debugging purposes." OFF)
# https://www.intel.com/content/www/us/en/develop/documentation/oneapi-dpcpp-cpp-compiler-dev-guide-and-reference/top/compilation/ahead-of-time-compilation.html
# acm-g10 is the target for the first Intel Arc Alchemist GPUs.
set(CYCLES_ONEAPI_SPIR64_GEN_DEVICES "acm-g10" CACHE STRING "oneAPI Intel GPU architectures to build binaries for")
set(CYCLES_ONEAPI_SYCL_TARGETS spir64 spir64_gen CACHE STRING "oneAPI targets to build AOT binaries for")
mark_as_advanced(WITH_CYCLES_ONEAPI_HOST_TASK_EXECUTION)
mark_as_advanced(CYCLES_ONEAPI_SPIR64_GEN_DEVICES)
mark_as_advanced(CYCLES_ONEAPI_SYCL_TARGETS)
endif()
@@ -597,8 +576,8 @@ endif()
# OpenGL
option(WITH_OPENGL "When off limits visibility of the opengl headers to just bf_gpu and gawain (temporary option for development purposes)" ON)
option(WITH_GPU_BUILDTIME_SHADER_BUILDER "Shader builder is a developer option enabling linting on GLSL during compilation" OFF)
option(WITH_OPENGL "When off limits visibility of the opengl headers to just bf_gpu and gawain (temporary option for development purposes)" ON)
option(WITH_GPU_BUILDTIME_SHADER_BUILDER "Shader builder is a developer option enabling linting on GLSL during compilation" OFF)
mark_as_advanced(
WITH_OPENGL
@@ -614,12 +593,16 @@ mark_as_advanced(
# Metal
if(APPLE)
option(WITH_METAL_BACKEND "Use Metal for graphics instead of (or as well as) OpenGL on macOS." ON)
option(WITH_METAL_BACKEND "Use Metal for graphics instead of (or as well as) OpenGL on macOS." OFF)
mark_as_advanced(WITH_METAL_BACKEND)
else()
set(WITH_METAL_BACKEND OFF)
endif()
if(WITH_METAL_BACKEND)
set(CMAKE_OSX_DEPLOYMENT_TARGET "10.15" CACHE STRING "Minimum OS X deployment version" FORCE)
endif()
if(WIN32)
getDefaultWindowsPrefixBase(CMAKE_GENERIC_PROGRAM_FILES)
set(CPACK_INSTALL_PREFIX ${CMAKE_GENERIC_PROGRAM_FILES}/${})
@@ -847,17 +830,27 @@ endif()
# enable boost for cycles, audaspace or i18n
# otherwise if the user disabled
set_and_warn_dependency(WITH_BOOST WITH_CYCLES OFF)
set_and_warn_dependency(WITH_BOOST WITH_INTERNATIONAL OFF)
set_and_warn_dependency(WITH_BOOST WITH_OPENVDB OFF)
set_and_warn_dependency(WITH_BOOST WITH_OPENCOLORIO OFF)
set_and_warn_dependency(WITH_BOOST WITH_QUADRIFLOW OFF)
set_and_warn_dependency(WITH_BOOST WITH_USD OFF)
set_and_warn_dependency(WITH_BOOST WITH_ALEMBIC OFF)
if(WITH_CYCLES)
set_and_warn_dependency(WITH_BOOST WITH_CYCLES_OSL OFF)
set_and_warn_dependency(WITH_PUGIXML WITH_CYCLES_OSL OFF)
endif()
set_and_warn_dependency(WITH_PUGIXML WITH_OPENIMAGEIO OFF)
if(WITH_BOOST AND NOT (WITH_CYCLES OR WITH_OPENIMAGEIO OR WITH_INTERNATIONAL OR
WITH_OPENVDB OR WITH_OPENCOLORIO OR WITH_USD OR WITH_ALEMBIC))
message(STATUS "No dependencies need 'WITH_BOOST' forcing WITH_BOOST=OFF")
set(WITH_BOOST OFF)
endif()
set_and_warn_dependency(WITH_TBB WITH_CYCLES OFF)
set_and_warn_dependency(WITH_TBB WITH_USD OFF)
set_and_warn_dependency(WITH_TBB WITH_OPENIMAGEDENOISE OFF)
set_and_warn_dependency(WITH_TBB WITH_OPENVDB OFF)
set_and_warn_dependency(WITH_TBB WITH_MOD_FLUID OFF)
@@ -866,10 +859,14 @@ set_and_warn_dependency(WITH_OPENVDB WITH_NANOVDB OFF)
# OpenVDB and OpenColorIO uses 'half' type from OpenEXR
set_and_warn_dependency(WITH_IMAGE_OPENEXR WITH_OPENVDB OFF)
set_and_warn_dependency(WITH_IMAGE_OPENEXR WITH_OPENCOLORIO OFF)
# Haru needs `TIFFFaxBlackCodes` & `TIFFFaxWhiteCodes` symbols from TIFF.
set_and_warn_dependency(WITH_IMAGE_TIFF WITH_HARU OFF)
# USD needs OpenSubDiv, since that is used by the Cycles Hydra render delegate.
set_and_warn_dependency(WITH_OPENSUBDIV WITH_USD OFF)
# auto enable openimageio for cycles
if(WITH_CYCLES)
set(WITH_OPENIMAGEIO ON)
@@ -883,6 +880,17 @@ else()
set(WITH_CYCLES_OSL OFF)
endif()
# auto enable openimageio linking dependencies
if(WITH_OPENIMAGEIO)
set(WITH_IMAGE_OPENEXR ON)
set(WITH_IMAGE_TIFF ON)
endif()
# auto enable alembic linking dependencies
if(WITH_ALEMBIC)
set(WITH_IMAGE_OPENEXR ON)
endif()
# don't store paths to libs for portable distribution
if(WITH_INSTALL_PORTABLE)
set(CMAKE_SKIP_BUILD_RPATH TRUE)
@@ -1008,8 +1016,6 @@ set(PLATFORM_LINKLIBS "")
# - CMAKE_EXE_LINKER_FLAGS_DEBUG
set(PLATFORM_LINKFLAGS "")
set(PLATFORM_LINKFLAGS_DEBUG "")
set(PLATFORM_LINKFLAGS_RELEASE "")
set(PLATFORM_LINKFLAGS_EXECUTABLE "")
if(NOT CMAKE_BUILD_TYPE MATCHES "Release")
if(WITH_COMPILER_ASAN)
@@ -1087,6 +1093,14 @@ if(WITH_CYCLES)
"Configure OIIO or disable WITH_CYCLES"
)
endif()
if(NOT WITH_BOOST)
message(
FATAL_ERROR
"Cycles requires WITH_BOOST, the library may not have been found. "
"Configure BOOST or disable WITH_CYCLES"
)
endif()
if(WITH_CYCLES_OSL)
if(NOT WITH_LLVM)
message(
@@ -1279,14 +1293,12 @@ endif()
# -----------------------------------------------------------------------------
# Configure Bullet
if(WITH_BULLET)
if(WITH_SYSTEM_BULLET)
find_package(Bullet)
set_and_warn_library_found("Bullet" BULLET_FOUND WITH_BULLET)
else()
set(BULLET_INCLUDE_DIRS "${CMAKE_SOURCE_DIR}/extern/bullet2/src")
set(BULLET_LIBRARIES "extern_bullet")
endif()
if(WITH_BULLET AND WITH_SYSTEM_BULLET)
find_package(Bullet)
set_and_warn_library_found("Bullet" BULLET_FOUND WITH_BULLET)
else()
set(BULLET_INCLUDE_DIRS "${CMAKE_SOURCE_DIR}/extern/bullet2/src")
# set(BULLET_LIBRARIES "")
endif()
@@ -1446,9 +1458,6 @@ if(CMAKE_COMPILER_IS_GNUCC)
add_check_c_compiler_flag(C_WARNINGS C_WARN_TYPE_LIMITS -Wtype-limits)
add_check_c_compiler_flag(C_WARNINGS C_WARN_FORMAT_SIGN -Wformat-signedness)
add_check_c_compiler_flag(C_WARNINGS C_WARN_RESTRICT -Wrestrict)
# Useful but too many false positives and inconvenient to suppress each occurrence.
add_check_c_compiler_flag(C_WARNINGS C_WARN_NO_STRINGOP_OVERREAD -Wno-stringop-overread)
add_check_c_compiler_flag(C_WARNINGS C_WARN_NO_STRINGOP_OVERFLOW -Wno-stringop-overflow)
# C-only.
add_check_c_compiler_flag(C_WARNINGS C_WARN_NO_NULL -Wnonnull)
@@ -1488,9 +1497,6 @@ if(CMAKE_COMPILER_IS_GNUCC)
add_check_cxx_compiler_flag(CXX_WARNINGS CXX_WARN_RESTRICT -Wrestrict)
add_check_cxx_compiler_flag(CXX_WARNINGS CXX_WARN_NO_SUGGEST_OVERRIDE -Wno-suggest-override)
add_check_cxx_compiler_flag(CXX_WARNINGS CXX_WARN_UNINITIALIZED -Wuninitialized)
# Useful but too many false positives and inconvenient to suppress each occurrence.
add_check_cxx_compiler_flag(CXX_WARNINGS CXX_WARN_NO_STRINGOP_OVERREAD -Wno-stringop-overread)
add_check_cxx_compiler_flag(CXX_WARNINGS CXX_WARN_NO_STRINGOP_OVERFLOW -Wno-stringop-overflow)
# causes too many warnings
if(NOT APPLE)

View File

@@ -1,38 +0,0 @@
<!--
Keep this document short & concise,
linking to external resources instead of including content in-line.
See 'release/text/readme.html' for the end user read-me.
-->
Blender
=======
Blender is the free and open source 3D creation suite.
It supports the entirety of the 3D pipeline-modeling, rigging, animation, simulation, rendering, compositing,
motion tracking and video editing.
![Blender screenshot](https://code.blender.org/wp-content/uploads/2018/12/springrg.jpg "Blender screenshot")
Project Pages
-------------
- [Main Website](http://www.blender.org)
- [Reference Manual](https://docs.blender.org/manual/en/latest/index.html)
- [User Community](https://www.blender.org/community/)
Development
-----------
- [Build Instructions](https://wiki.blender.org/wiki/Building_Blender)
- [Code Review & Bug Tracker](https://developer.blender.org)
- [Developer Forum](https://devtalk.blender.org)
- [Developer Documentation](https://wiki.blender.org)
License
-------
Blender as a whole is licensed under the GNU General Public License, Version 3.
Individual files may have a different, but compatible license.
See [blender.org/about/license](https://www.blender.org/about/license) for details.

View File

@@ -24,7 +24,7 @@
set(SBOMCONTENTS)
get_cmake_property(_variableNames VARIABLES)
foreach(_variableName ${_variableNames})
foreach (_variableName ${_variableNames})
if(_variableName MATCHES "CPE$")
string(REPLACE ":" ";" CPE_LIST ${${_variableName}})
string(REPLACE "_CPE" "_ID" CPE_DEPNAME ${_variableName})

View File

@@ -10,22 +10,22 @@ function(download_source dep)
if(PACKAGE_USE_UPSTREAM_SOURCES)
set(TARGET_URI ${${dep}_URI})
elseif(BLENDER_VERSION)
set(TARGET_URI https://svn.blender.org/svnroot/bf-blender/tags/blender-${BLENDER_VERSION}-release/lib/packages/${TARGET_FILE})
set(TARGET_URI https://svn.blender.org/svnroot/bf-blender/tags/blender-${BLENDER_VERSION}-release/lib/packages/${TARGET_FILE})
else()
set(TARGET_URI https://svn.blender.org/svnroot/bf-blender/trunk/lib/packages/${TARGET_FILE})
set(TARGET_URI https://svn.blender.org/svnroot/bf-blender/trunk/lib/packages/${TARGET_FILE})
endif()
# Validate all required variables are set and give an explicit error message
# rather than CMake erroring out later on with a more ambigious error.
if(NOT DEFINED TARGET_FILE)
if (NOT DEFINED TARGET_FILE)
message(FATAL_ERROR "${dep}_FILE variable not set")
endif()
if(NOT DEFINED TARGET_HASH)
if (NOT DEFINED TARGET_HASH)
message(FATAL_ERROR "${dep}_HASH variable not set")
endif()
if(NOT DEFINED TARGET_HASH_TYPE)
if (NOT DEFINED TARGET_HASH_TYPE)
message(FATAL_ERROR "${dep}_HASH_TYPE variable not set")
endif()
if(NOT DEFINED TARGET_URI)
if (NOT DEFINED TARGET_URI)
message(FATAL_ERROR "${dep}_URI variable not set")
endif()
set(TARGET_FILE ${PACKAGE_DIR}/${TARGET_FILE})
@@ -43,11 +43,11 @@ function(download_source dep)
if(EXISTS ${TARGET_FILE})
# Sometimes the download fails, but that is not a
# fail condition for "file(DOWNLOAD" it will warn about
# a CRC mismatch and just carry on, we need to explicitly
# a crc mismatch and just carry on, we need to explicitly
# catch this and remove the bogus 0 byte file so we can
# retry without having to go find the file and manually
# delete it.
file(SIZE ${TARGET_FILE} TARGET_SIZE)
file (SIZE ${TARGET_FILE} TARGET_SIZE)
if(${TARGET_SIZE} EQUAL 0)
file(REMOVE ${TARGET_FILE})
message(FATAL_ERROR "for ${TARGET_FILE} file size 0, download likely failed, deleted...")

View File

@@ -15,11 +15,10 @@ set(EMBREE_EXTRA_ARGS
-DTBB_ROOT=${LIBDIR}/tbb
)
if(NOT BLENDER_PLATFORM_ARM)
if (NOT BLENDER_PLATFORM_ARM)
set(EMBREE_EXTRA_ARGS
${EMBREE_EXTRA_ARGS}
-DEMBREE_MAX_ISA=AVX2
)
-DEMBREE_MAX_ISA=AVX2)
endif()
if(TBB_STATIC_LIBRARY)
@@ -53,17 +52,17 @@ if(WIN32)
)
else()
ExternalProject_Add_Step(external_embree after_install
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree3.lib ${HARVEST_TARGET}/embree/lib/embree3_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_avx.lib ${HARVEST_TARGET}/embree/lib/embree_avx_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_avx2.lib ${HARVEST_TARGET}/embree/lib/embree_avx2_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_sse42.lib ${HARVEST_TARGET}/embree/lib/embree_sse42_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/lexers.lib ${HARVEST_TARGET}/embree/lib/lexers_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/math.lib ${HARVEST_TARGET}/embree/lib/math_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/simd.lib ${HARVEST_TARGET}/embree/lib/simd_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/sys.lib ${HARVEST_TARGET}/embree/lib/sys_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/tasking.lib ${HARVEST_TARGET}/embree/lib/tasking_d.lib
DEPENDEES install
)
endif()
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree3.lib ${HARVEST_TARGET}/embree/lib/embree3_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_avx.lib ${HARVEST_TARGET}/embree/lib/embree_avx_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_avx2.lib ${HARVEST_TARGET}/embree/lib/embree_avx2_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/embree_sse42.lib ${HARVEST_TARGET}/embree/lib/embree_sse42_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/lexers.lib ${HARVEST_TARGET}/embree/lib/lexers_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/math.lib ${HARVEST_TARGET}/embree/lib/math_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/simd.lib ${HARVEST_TARGET}/embree/lib/simd_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/sys.lib ${HARVEST_TARGET}/embree/lib/sys_d.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/embree/lib/tasking.lib ${HARVEST_TARGET}/embree/lib/tasking_d.lib
DEPENDEES install
)
endif()
endif()

View File

@@ -22,14 +22,6 @@ elseif(UNIX AND NOT APPLE)
)
endif()
# Boolean crashes with Arm assembly, see T103423.
if(BLENDER_PLATFORM_ARM)
set(GMP_OPTIONS
${GMP_OPTIONS}
--disable-assembly
)
endif()
ExternalProject_Add(external_gmp
URL file://${PACKAGE_DIR}/${GMP_FILE}
DOWNLOAD_DIR ${DOWNLOAD_DIR}
@@ -48,10 +40,10 @@ endif()
if(BUILD_MODE STREQUAL Release AND WIN32)
ExternalProject_Add_Step(external_gmp after_install
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-3.dll.def ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def
COMMAND lib /def:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def /machine:x64 /out:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/gmp/bin/libgmp-10.dll ${HARVEST_TARGET}/gmp/lib/libgmp-10.dll
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib ${HARVEST_TARGET}/gmp/lib/libgmp-10.lib
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-3.dll.def ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def
COMMAND lib /def:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def /machine:x64 /out:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib
COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/gmp/bin/libgmp-10.dll ${HARVEST_TARGET}/gmp/lib/libgmp-10.dll
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib ${HARVEST_TARGET}/gmp/lib/libgmp-10.lib
COMMAND ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/gmp/include ${HARVEST_TARGET}/gmp/include
DEPENDEES install
@@ -60,8 +52,8 @@ endif()
if(BUILD_MODE STREQUAL Debug AND WIN32)
ExternalProject_Add_Step(external_gmp after_install
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-3.dll.def ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def
COMMAND lib /def:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def /machine:x64 /out:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib
COMMAND ${CMAKE_COMMAND} -E copy ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-3.dll.def ${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def
COMMAND lib /def:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.def /machine:x64 /out:${BUILD_DIR}/gmp/src/external_gmp/.libs/libgmp-10.lib
DEPENDEES install
)

View File

@@ -11,301 +11,274 @@ message("HARVEST_TARGET = ${HARVEST_TARGET}")
if(WIN32)
if(BUILD_MODE STREQUAL Release)
add_custom_target(Harvest_Release_Results
COMMAND # JPEG rename lib-file + copy include.
if(BUILD_MODE STREQUAL Release)
add_custom_target(Harvest_Release_Results
COMMAND # jpeg rename libfile + copy include
${CMAKE_COMMAND} -E copy ${LIBDIR}/jpeg/lib/jpeg-static.lib ${HARVEST_TARGET}/jpeg/lib/libjpeg.lib &&
${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/jpeg/include/ ${HARVEST_TARGET}/jpeg/include/ &&
# PNG.
# png
${CMAKE_COMMAND} -E copy ${LIBDIR}/png/lib/libpng16_static.lib ${HARVEST_TARGET}/png/lib/libpng.lib &&
${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/png/include/ ${HARVEST_TARGET}/png/include/ &&
# FREEGLUT -> OPENGL.
# freeglut-> opengl
${CMAKE_COMMAND} -E copy ${LIBDIR}/freeglut/lib/freeglut_static.lib ${HARVEST_TARGET}/opengl/lib/freeglut_static.lib &&
${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/freeglut/include/ ${HARVEST_TARGET}/opengl/include/ &&
DEPENDS
)
endif()
DEPENDS
else(WIN32)
function(harvest from to)
set(pattern "")
foreach(f ${ARGN})
set(pattern ${f})
endforeach()
if(pattern STREQUAL "")
get_filename_component(dirpath ${to} DIRECTORY)
get_filename_component(filename ${to} NAME)
install(
FILES ${LIBDIR}/${from}
DESTINATION ${HARVEST_TARGET}/${dirpath}
RENAME ${filename}
)
endif()
else()
function(harvest from to)
set(pattern "")
foreach(f ${ARGN})
set(pattern ${f})
endforeach()
if(pattern STREQUAL "")
get_filename_component(dirpath ${to} DIRECTORY)
get_filename_component(filename ${to} NAME)
install(
FILES ${LIBDIR}/${from}
DESTINATION ${HARVEST_TARGET}/${dirpath}
RENAME ${filename}
)
else()
install(
DIRECTORY ${LIBDIR}/${from}/
DESTINATION ${HARVEST_TARGET}/${to}
USE_SOURCE_PERMISSIONS
FILES_MATCHING PATTERN ${pattern}
PATTERN "pkgconfig" EXCLUDE
PATTERN "cmake" EXCLUDE
PATTERN "__pycache__" EXCLUDE
PATTERN "tests" EXCLUDE
PATTERN "meson*" EXCLUDE
)
endif()
endfunction()
# Set rpath on shared libraries to $ORIGIN since all will be installed in the same
# lib folder, and remove any absolute paths.
#
# Ideally this would be done as part of the Blender build since it makes assumptions
# about where the files will be installed. However it would add patchelf as a new
# dependency for building.
#
# Also removes versioned symlinks, which give errors with macOS notarization.
if(APPLE)
set(set_rpath_cmd python3 ${CMAKE_CURRENT_SOURCE_DIR}/darwin/set_rpath.py @loader_path)
else()
set(set_rpath_cmd patchelf --set-rpath $ORIGIN)
endif()
function(harvest_rpath_lib from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
cmake_policy(SET CMP0009 NEW)\n
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}) \n
foreach(f \${shared_libs}) \n
if(IS_SYMLINK \${f})\n
if(APPLE)\n
file(REMOVE_RECURSE \${f})
endif()\n
else()\n
execute_process(COMMAND ${set_rpath_cmd} \${f}) \n
endif()\n
endforeach()")
endfunction()
# Set rpath on utility binaries assuming they are run from their install location.
function(harvest_rpath_bin from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}) \n
foreach(f \${shared_libs}) \n
execute_process(COMMAND ${set_rpath_cmd}/../lib; \${f}) \n
endforeach()")
endfunction()
# Set rpath on Python module to point to the shared libraries folder in the Blender
# installation.
function(harvest_rpath_python from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}\.so*) \n
foreach(f \${shared_libs}) \n
if(IS_SYMLINK \${f})\n
if(APPLE)\n
file(REMOVE_RECURSE \${f})
endif()\n
else()\n
get_filename_component(f_dir \${f} DIRECTORY) \n
file(RELATIVE_PATH relative_dir \${f_dir} ${HARVEST_TARGET}) \n
execute_process(COMMAND ${set_rpath_cmd}/\${relative_dir}../lib \${f}) \n
endif()\n
endforeach()")
endfunction()
harvest(alembic/include alembic/include "*.h")
harvest(alembic/lib/libAlembic.a alembic/lib/libAlembic.a)
harvest_rpath_bin(alembic/bin alembic/bin "*")
harvest(brotli/include brotli/include "*.h")
harvest(brotli/lib brotli/lib "*.a")
harvest(boost/include boost/include "*")
harvest_rpath_lib(boost/lib boost/lib "*${SHAREDLIBEXT}*")
harvest(imath/include imath/include "*.h")
harvest_rpath_lib(imath/lib imath/lib "*${SHAREDLIBEXT}*")
harvest(ffmpeg/include ffmpeg/include "*.h")
harvest(ffmpeg/lib ffmpeg/lib "*.a")
harvest(fftw3/include fftw3/include "*.h")
harvest(fftw3/lib fftw3/lib "*.a")
harvest(flac/lib sndfile/lib "libFLAC.a")
harvest(freetype/include freetype/include "*.h")
harvest(freetype/lib/libfreetype2ST.a freetype/lib/libfreetype.a)
harvest(fribidi/include fribidi/include "*.h")
harvest(fribidi/lib fribidi/lib "*.a")
harvest(epoxy/include epoxy/include "*.h")
harvest(epoxy/lib epoxy/lib "*.a")
harvest(gmp/include gmp/include "*.h")
harvest(gmp/lib gmp/lib "*.a")
harvest(harfbuzz/include harfbuzz/include "*.h")
harvest(harfbuzz/lib harfbuzz/lib "*.a")
harvest(jemalloc/include jemalloc/include "*.h")
harvest(jemalloc/lib jemalloc/lib "*.a")
harvest(jpeg/include jpeg/include "*.h")
harvest(jpeg/lib jpeg/lib "libjpeg.a")
harvest(lame/lib ffmpeg/lib "*.a")
if(NOT APPLE)
harvest(level-zero/include/level_zero level-zero/include/level_zero "*.h")
harvest(level-zero/lib level-zero/lib "*${SHAREDLIBEXT}*")
endif()
harvest(llvm/bin llvm/bin "clang-format")
if(BUILD_CLANG_TOOLS)
harvest(llvm/bin llvm/bin "clang-tidy")
harvest(llvm/share/clang llvm/share "run-clang-tidy.py")
endif()
harvest(llvm/include llvm/include "*")
harvest(llvm/bin llvm/bin "llvm-config")
harvest(llvm/lib llvm/lib "libLLVM*.a")
harvest(llvm/lib llvm/lib "libclang*.a")
harvest(llvm/lib/clang llvm/lib/clang "*.h")
if(APPLE)
harvest(openmp/lib openmp/lib "libomp.dylib")
harvest(openmp/include openmp/include "*.h")
endif()
if(BLENDER_PLATFORM_ARM)
harvest(sse2neon sse2neon "*.h")
endif()
harvest(ogg/lib ffmpeg/lib "*.a")
harvest(openal/include openal/include "*.h")
if(UNIX AND NOT APPLE)
harvest(openal/lib openal/lib "*.a")
harvest(zlib/include zlib/include "*.h")
harvest(zlib/lib zlib/lib "*.a")
harvest(xml2/include xml2/include "*.h")
harvest(xml2/lib xml2/lib "*.a")
harvest(
wayland-protocols/share/wayland-protocols
wayland-protocols/share/wayland-protocols/
"*.xml"
install(
DIRECTORY ${LIBDIR}/${from}/
DESTINATION ${HARVEST_TARGET}/${to}
USE_SOURCE_PERMISSIONS
FILES_MATCHING PATTERN ${pattern}
PATTERN "pkgconfig" EXCLUDE
PATTERN "cmake" EXCLUDE
PATTERN "__pycache__" EXCLUDE
PATTERN "tests" EXCLUDE
PATTERN "meson*" EXCLUDE
)
endif()
endfunction()
# Set rpath on shared libraries to $ORIGIN since all will be installed in the same
# lib folder, and remove any absolute paths.
#
# Ideally this would be done as part of the Blender build since it makes assumptions
# about where the files will be installed. However it would add patchelf as a new
# dependency for building.
#
# Also removes versioned symlinks, which give errors with macOS notarization.
if(APPLE)
set(set_rpath_cmd python3 ${CMAKE_CURRENT_SOURCE_DIR}/darwin/set_rpath.py @loader_path)
else()
set(set_rpath_cmd patchelf --set-rpath $ORIGIN)
endif()
function(harvest_rpath_lib from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
cmake_policy(SET CMP0009 NEW)\n
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}) \n
foreach(f \${shared_libs}) \n
if(IS_SYMLINK \${f})\n
if(APPLE)\n
file(REMOVE_RECURSE \${f})
endif()\n
else()\n
execute_process(COMMAND ${set_rpath_cmd} \${f}) \n
endif()\n
endforeach()")
endfunction()
# Set rpath on utility binaries assuming they are run from their install location.
function(harvest_rpath_bin from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}) \n
foreach(f \${shared_libs}) \n
execute_process(COMMAND ${set_rpath_cmd}/../lib; \${f}) \n
endforeach()")
endfunction()
# Set rpath on Python module to point to the shared libraries folder in the Blender
# installation.
function(harvest_rpath_python from to pattern)
harvest(${from} ${to} ${pattern})
install(CODE "\
file(GLOB_RECURSE shared_libs ${HARVEST_TARGET}/${to}/${pattern}\.so*) \n
foreach(f \${shared_libs}) \n
if(IS_SYMLINK \${f})\n
if(APPLE)\n
file(REMOVE_RECURSE \${f})
endif()\n
else()\n
get_filename_component(f_dir \${f} DIRECTORY) \n
file(RELATIVE_PATH relative_dir \${f_dir} ${HARVEST_TARGET}) \n
execute_process(COMMAND ${set_rpath_cmd}/\${relative_dir}../lib \${f}) \n
endif()\n
endforeach()")
endfunction()
harvest(alembic/include alembic/include "*.h")
harvest(alembic/lib/libAlembic.a alembic/lib/libAlembic.a)
harvest_rpath_bin(alembic/bin alembic/bin "*")
harvest(brotli/include brotli/include "*.h")
harvest(brotli/lib brotli/lib "*.a")
harvest(boost/include boost/include "*")
harvest_rpath_lib(boost/lib boost/lib "*${SHAREDLIBEXT}*")
harvest(imath/include imath/include "*.h")
harvest_rpath_lib(imath/lib imath/lib "*${SHAREDLIBEXT}*")
harvest(ffmpeg/include ffmpeg/include "*.h")
harvest(ffmpeg/lib ffmpeg/lib "*.a")
harvest(fftw3/include fftw3/include "*.h")
harvest(fftw3/lib fftw3/lib "*.a")
harvest(flac/lib sndfile/lib "libFLAC.a")
harvest(freetype/include freetype/include "*.h")
harvest(freetype/lib/libfreetype2ST.a freetype/lib/libfreetype.a)
harvest(fribidi/include fribidi/include "*.h")
harvest(fribidi/lib fribidi/lib "*.a")
harvest(epoxy/include epoxy/include "*.h")
harvest(epoxy/lib epoxy/lib "*.a")
harvest(gmp/include gmp/include "*.h")
harvest(gmp/lib gmp/lib "*.a")
harvest(harfbuzz/include harfbuzz/include "*.h")
harvest(harfbuzz/lib harfbuzz/lib "*.a")
harvest(jemalloc/include jemalloc/include "*.h")
harvest(jemalloc/lib jemalloc/lib "*.a")
harvest(jpeg/include jpeg/include "*.h")
harvest(jpeg/lib jpeg/lib "libjpeg.a")
harvest(lame/lib ffmpeg/lib "*.a")
if(NOT APPLE)
harvest(level-zero/include/level_zero level-zero/include/level_zero "*.h")
harvest(level-zero/lib level-zero/lib "*${SHAREDLIBEXT}*")
endif()
harvest(llvm/bin llvm/bin "clang-format")
if(BUILD_CLANG_TOOLS)
harvest(llvm/bin llvm/bin "clang-tidy")
harvest(llvm/share/clang llvm/share "run-clang-tidy.py")
endif()
harvest(llvm/include llvm/include "*")
harvest(llvm/bin llvm/bin "llvm-config")
harvest(llvm/lib llvm/lib "libLLVM*.a")
harvest(llvm/lib llvm/lib "libclang*.a")
harvest(llvm/lib/clang llvm/lib/clang "*.h")
if(APPLE)
harvest(openmp/lib openmp/lib "libomp.dylib")
harvest(openmp/include openmp/include "*.h")
endif()
if(BLENDER_PLATFORM_ARM)
harvest(sse2neon sse2neon "*.h")
endif()
harvest(ogg/lib ffmpeg/lib "*.a")
harvest(openal/include openal/include "*.h")
if(UNIX AND NOT APPLE)
harvest(openal/lib openal/lib "*.a")
harvest(zlib/include zlib/include "*.h")
harvest(zlib/lib zlib/lib "*.a")
harvest(xml2/include xml2/include "*.h")
harvest(xml2/lib xml2/lib "*.a")
harvest(wayland-protocols/share/wayland-protocols wayland-protocols/share/wayland-protocols/ "*.xml")
harvest(wayland/bin wayland/bin "wayland-scanner")
harvest(wayland/include wayland/include "*.h")
harvest(wayland_libdecor/include wayland_libdecor/include "*.h")
else()
harvest(blosc/lib openvdb/lib "*.a")
harvest(xml2/lib opencollada/lib "*.a")
endif()
harvest(opencollada/include/opencollada opencollada/include "*.h")
harvest(opencollada/lib/opencollada opencollada/lib "*.a")
harvest(opencolorio/include opencolorio/include "*.h")
harvest_rpath_lib(opencolorio/lib opencolorio/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(
opencolorio/lib/python${PYTHON_SHORT_VERSION}
python/lib/python${PYTHON_SHORT_VERSION}
"*"
)
harvest(openexr/include openexr/include "*.h")
harvest_rpath_lib(openexr/lib openexr/lib "*${SHAREDLIBEXT}*")
harvest_rpath_bin(openimageio/bin openimageio/bin "idiff")
harvest_rpath_bin(openimageio/bin openimageio/bin "maketx")
harvest_rpath_bin(openimageio/bin openimageio/bin "oiiotool")
harvest(openimageio/include openimageio/include "*")
harvest_rpath_lib(openimageio/lib openimageio/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(
openimageio/lib/python${PYTHON_SHORT_VERSION}
python/lib/python${PYTHON_SHORT_VERSION}
"*"
)
harvest(openimagedenoise/include openimagedenoise/include "*")
harvest(openimagedenoise/lib openimagedenoise/lib "*.a")
harvest(embree/include embree/include "*.h")
harvest(embree/lib embree/lib "*.a")
harvest(openpgl/include openpgl/include "*.h")
harvest(openpgl/lib openpgl/lib "*.a")
harvest(openpgl/lib/cmake/openpgl-${OPENPGL_SHORT_VERSION} openpgl/lib/cmake/openpgl "*.cmake")
harvest(openjpeg/include/openjpeg-${OPENJPEG_SHORT_VERSION} openjpeg/include "*.h")
harvest(openjpeg/lib openjpeg/lib "*.a")
harvest(opensubdiv/include opensubdiv/include "*.h")
harvest_rpath_lib(opensubdiv/lib opensubdiv/lib "*${SHAREDLIBEXT}*")
harvest(openvdb/include/openvdb openvdb/include/openvdb "*.h")
harvest(openvdb/include/nanovdb openvdb/include/nanovdb "*.h")
harvest_rpath_lib(openvdb/lib openvdb/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(
openvdb/lib/python${PYTHON_SHORT_VERSION}
python/lib/python${PYTHON_SHORT_VERSION}
"*pyopenvdb*"
)
harvest(xr_openxr_sdk/include/openxr xr_openxr_sdk/include/openxr "*.h")
harvest(xr_openxr_sdk/lib xr_openxr_sdk/lib "*.a")
harvest_rpath_bin(osl/bin osl/bin "oslc")
harvest(osl/include osl/include "*.h")
harvest(osl/lib osl/lib "*.a")
harvest(osl/share/OSL/shaders osl/share/OSL/shaders "*.h")
harvest(png/include png/include "*.h")
harvest(png/lib png/lib "*.a")
harvest(pugixml/include pugixml/include "*.hpp")
harvest(pugixml/lib pugixml/lib "*.a")
harvest(python/bin python/bin "python${PYTHON_SHORT_VERSION}")
harvest(python/include python/include "*h")
harvest(python/lib python/lib "*")
harvest(sdl/include/SDL2 sdl/include "*.h")
harvest(sdl/lib sdl/lib "libSDL2.a")
harvest(sndfile/include sndfile/include "*.h")
harvest(sndfile/lib sndfile/lib "*.a")
harvest(spnav/include spnav/include "*.h")
harvest(spnav/lib spnav/lib "*.a")
harvest(tbb/include tbb/include "*.h")
harvest_rpath_lib(tbb/lib tbb/lib "libtbb${SHAREDLIBEXT}")
harvest(theora/lib ffmpeg/lib "*.a")
harvest(tiff/include tiff/include "*.h")
harvest(tiff/lib tiff/lib "*.a")
harvest(vorbis/lib ffmpeg/lib "*.a")
harvest(opus/lib ffmpeg/lib "*.a")
harvest(vpx/lib ffmpeg/lib "*.a")
harvest(x264/lib ffmpeg/lib "*.a")
harvest(xvidcore/lib ffmpeg/lib "*.a")
harvest(aom/lib ffmpeg/lib "*.a")
harvest(webp/lib webp/lib "*.a")
harvest(webp/include webp/include "*.h")
harvest(usd/include usd/include "*.h")
harvest_rpath_lib(usd/lib usd/lib "libusd_ms${SHAREDLIBEXT}")
harvest(usd/lib/usd usd/lib/usd "*")
harvest_rpath_python(
usd/lib/python/pxr
python/lib/python${PYTHON_SHORT_VERSION}/site-packages/pxr
"*"
)
harvest(usd/plugin usd/plugin "*")
harvest(materialx/include materialx/include "*.h")
harvest_rpath_lib(materialx/lib materialx/lib "*${SHAREDLIBEXT}*")
harvest(materialx/libraries materialx/libraries "*")
harvest(materialx/lib/cmake/MaterialX materialx/lib/cmake/MaterialX "*.cmake")
harvest_rpath_python(
materialx/python/MaterialX
python/lib/python${PYTHON_SHORT_VERSION}/site-packages/MaterialX
"*"
)
# We do not need anything from the resources folder, but the MaterialX config
# file will complain if the folder does not exist, so just copy the readme.md
# files to ensure the folder will exist.
harvest(materialx/resources materialx/resources "README.md")
harvest(potrace/include potrace/include "*.h")
harvest(potrace/lib potrace/lib "*.a")
harvest(haru/include haru/include "*.h")
harvest(haru/lib haru/lib "*.a")
harvest(zstd/include zstd/include "*.h")
harvest(zstd/lib zstd/lib "*.a")
harvest(shaderc shaderc "*")
harvest(vulkan_headers vulkan "*")
harvest_rpath_lib(vulkan_loader/lib vulkan/lib "*${SHAREDLIBEXT}*")
if(APPLE)
harvest(vulkan_loader/loader vulkan/loader "*")
endif()
if(UNIX AND NOT APPLE)
harvest(libglu/lib mesa/lib "*${SHAREDLIBEXT}*")
harvest(mesa/lib64 mesa/lib "*${SHAREDLIBEXT}*")
harvest(dpcpp dpcpp "*")
harvest(igc dpcpp/lib/igc "*")
harvest(ocloc dpcpp/lib/ocloc "*")
endif()
harvest(xml2/lib opencollada/lib "*.a")
endif()
harvest(opencollada/include/opencollada opencollada/include "*.h")
harvest(opencollada/lib/opencollada opencollada/lib "*.a")
harvest(opencolorio/include opencolorio/include "*.h")
harvest_rpath_lib(opencolorio/lib opencolorio/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(opencolorio/lib/python${PYTHON_SHORT_VERSION} python/lib/python${PYTHON_SHORT_VERSION} "*")
harvest(openexr/include openexr/include "*.h")
harvest_rpath_lib(openexr/lib openexr/lib "*${SHAREDLIBEXT}*")
harvest_rpath_bin(openimageio/bin openimageio/bin "idiff")
harvest_rpath_bin(openimageio/bin openimageio/bin "maketx")
harvest_rpath_bin(openimageio/bin openimageio/bin "oiiotool")
harvest(openimageio/include openimageio/include "*")
harvest_rpath_lib(openimageio/lib openimageio/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(openimageio/lib/python${PYTHON_SHORT_VERSION} python/lib/python${PYTHON_SHORT_VERSION} "*")
harvest(openimagedenoise/include openimagedenoise/include "*")
harvest(openimagedenoise/lib openimagedenoise/lib "*.a")
harvest(embree/include embree/include "*.h")
harvest(embree/lib embree/lib "*.a")
harvest(openpgl/include openpgl/include "*.h")
harvest(openpgl/lib openpgl/lib "*.a")
harvest(openpgl/lib/cmake/openpgl-${OPENPGL_SHORT_VERSION} openpgl/lib/cmake/openpgl "*.cmake")
harvest(openjpeg/include/openjpeg-${OPENJPEG_SHORT_VERSION} openjpeg/include "*.h")
harvest(openjpeg/lib openjpeg/lib "*.a")
harvest(opensubdiv/include opensubdiv/include "*.h")
harvest_rpath_lib(opensubdiv/lib opensubdiv/lib "*${SHAREDLIBEXT}*")
harvest(openvdb/include/openvdb openvdb/include/openvdb "*.h")
harvest(openvdb/include/nanovdb openvdb/include/nanovdb "*.h")
harvest_rpath_lib(openvdb/lib openvdb/lib "*${SHAREDLIBEXT}*")
harvest_rpath_python(openvdb/lib/python${PYTHON_SHORT_VERSION} python/lib/python${PYTHON_SHORT_VERSION} "*pyopenvdb*")
harvest(xr_openxr_sdk/include/openxr xr_openxr_sdk/include/openxr "*.h")
harvest(xr_openxr_sdk/lib xr_openxr_sdk/lib "*.a")
harvest_rpath_bin(osl/bin osl/bin "oslc")
harvest(osl/include osl/include "*.h")
harvest(osl/lib osl/lib "*.a")
harvest(osl/share/OSL/shaders osl/share/OSL/shaders "*.h")
harvest(png/include png/include "*.h")
harvest(png/lib png/lib "*.a")
harvest(pugixml/include pugixml/include "*.hpp")
harvest(pugixml/lib pugixml/lib "*.a")
harvest(python/bin python/bin "python${PYTHON_SHORT_VERSION}")
harvest(python/include python/include "*h")
harvest(python/lib python/lib "*")
harvest(sdl/include/SDL2 sdl/include "*.h")
harvest(sdl/lib sdl/lib "libSDL2.a")
harvest(sndfile/include sndfile/include "*.h")
harvest(sndfile/lib sndfile/lib "*.a")
harvest(spnav/include spnav/include "*.h")
harvest(spnav/lib spnav/lib "*.a")
harvest(tbb/include tbb/include "*.h")
harvest_rpath_lib(tbb/lib tbb/lib "libtbb${SHAREDLIBEXT}")
harvest(theora/lib ffmpeg/lib "*.a")
harvest(tiff/include tiff/include "*.h")
harvest(tiff/lib tiff/lib "*.a")
harvest(vorbis/lib ffmpeg/lib "*.a")
harvest(opus/lib ffmpeg/lib "*.a")
harvest(vpx/lib ffmpeg/lib "*.a")
harvest(x264/lib ffmpeg/lib "*.a")
harvest(xvidcore/lib ffmpeg/lib "*.a")
harvest(aom/lib ffmpeg/lib "*.a")
harvest(webp/lib webp/lib "*.a")
harvest(webp/include webp/include "*.h")
harvest(usd/include usd/include "*.h")
harvest_rpath_lib(usd/lib usd/lib "libusd_ms${SHAREDLIBEXT}")
harvest(usd/lib/usd usd/lib/usd "*")
harvest_rpath_python(usd/lib/python/pxr python/lib/python${PYTHON_SHORT_VERSION}/site-packages/pxr "*")
harvest(usd/plugin usd/plugin "*")
harvest(materialx/include materialx/include "*.h")
harvest_rpath_lib(materialx/lib materialx/lib "*${SHAREDLIBEXT}*")
harvest(materialx/libraries materialx/libraries "*")
harvest(materialx/lib/cmake/MaterialX materialx/lib/cmake/MaterialX "*.cmake")
harvest_rpath_python(materialx/python/MaterialX python/lib/python${PYTHON_SHORT_VERSION}/site-packages/MaterialX "*")
# We do not need anything from the resources folder, but the MaterialX config
# file will complain if the folder does not exist, so just copy the readme.md
# files to ensure the folder will exist.
harvest(materialx/resources materialx/resources "README.md")
harvest(potrace/include potrace/include "*.h")
harvest(potrace/lib potrace/lib "*.a")
harvest(haru/include haru/include "*.h")
harvest(haru/lib haru/lib "*.a")
harvest(zstd/include zstd/include "*.h")
harvest(zstd/lib zstd/lib "*.a")
harvest(shaderc shaderc "*")
harvest(vulkan_headers vulkan "*")
harvest_rpath_lib(vulkan_loader/lib vulkan/lib "*${SHAREDLIBEXT}*")
harvest(vulkan_loader/loader vulkan/loader "*")
if(UNIX AND NOT APPLE)
harvest(libglu/lib mesa/lib "*${SHAREDLIBEXT}*")
harvest(mesa/lib64 mesa/lib "*${SHAREDLIBEXT}*")
harvest(dpcpp dpcpp "*")
harvest(igc dpcpp/lib/igc "*")
harvest(ocloc dpcpp/lib/ocloc "*")
endif()
endif()

View File

@@ -26,9 +26,9 @@ else()
-DLIBXML2_INCLUDE_DIR=${LIBDIR}/xml2/include/libxml2
)
if(BUILD_MODE STREQUAL Release)
list(APPEND OPENCOLLADA_EXTRA_ARGS -DLIBXML2_LIBRARIES=${LIBDIR}/xml2/lib/libxml2s.lib)
list(APPEND OPENCOLLADA_EXTRA_ARGS -DLIBXML2_LIBRARIES=${LIBDIR}/xml2/lib/libxml2s.lib)
else()
list(APPEND OPENCOLLADA_EXTRA_ARGS -DLIBXML2_LIBRARIES=${LIBDIR}/xml2/lib/libxml2sd.lib)
list(APPEND OPENCOLLADA_EXTRA_ARGS -DLIBXML2_LIBRARIES=${LIBDIR}/xml2/lib/libxml2sd.lib)
endif()
set(PATCH_MAYBE_DOS2UNIX_CMD)
endif()

View File

@@ -1,5 +1,20 @@
# SPDX-License-Identifier: GPL-2.0-or-later
# ***** BEGIN GPL LICENSE BLOCK *****
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software Foundation,
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ***** END GPL LICENSE BLOCK *****
set(SHADERC_EXTRA_ARGS
-DSHADERC_SKIP_TESTS=On
-DSHADERC_SPIRV_TOOLS_DIR=${BUILD_DIR}/shaderc_spirv_tools/src/external_shaderc_spirv_tools
@@ -45,3 +60,5 @@ if(WIN32)
)
endif()
endif()

View File

@@ -1,4 +1,20 @@
# SPDX-License-Identifier: GPL-2.0-or-later
# ***** BEGIN GPL LICENSE BLOCK *****
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software Foundation,
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ***** END GPL LICENSE BLOCK *****
# These are build time requirements for shaderc. We only have to unpack these
# shaderc will build them.
@@ -32,3 +48,4 @@ ExternalProject_Add(external_shaderc_spirv_tools
BUILD_COMMAND echo .
INSTALL_COMMAND echo .
)

View File

@@ -17,6 +17,19 @@ else()
set(SNDFILE_OPTIONS --enable-static --disable-shared )
endif()
if(UNIX AND NOT APPLE)
# NOTE(@campbellbarton): For some reason OPUS is alone in referencing the sub-directory,
# manipulate the package-config file to prevent this from happening.
# There is no problem with applying this change multiple times.
#
# Replace: Cflags: -I${includedir}/opus
# With: Cflags: -I${includedir}
set(SNDFILE_ENV
sed -i s/{includedir}\\/opus/{includedir}/g ${LIBDIR}/opus/lib/pkgconfig/opus.pc &&
${SNDFILE_ENV}
)
endif()
ExternalProject_Add(external_sndfile
URL file://${PACKAGE_DIR}/${SNDFILE_FILE}
DOWNLOAD_DIR ${DOWNLOAD_DIR}

View File

@@ -29,13 +29,12 @@ elseif(UNIX)
set(USD_PLATFORM_FLAGS
-DPYTHON_INCLUDE_DIR=${LIBDIR}/python/include/python${PYTHON_SHORT_VERSION}/
-DPYTHON_LIBRARY=${LIBDIR}/tbb/lib/${LIBPREFIX}${TBB_LIBRARY}${SHAREDLIBEXT}
)
)
if(APPLE)
set(USD_SHARED_LINKER_FLAGS "-Xlinker -undefined -Xlinker dynamic_lookup")
list(APPEND USD_PLATFORM_FLAGS
-DCMAKE_SHARED_LINKER_FLAGS=${USD_SHARED_LINKER_FLAGS}
)
-DCMAKE_SHARED_LINKER_FLAGS=${USD_SHARED_LINKER_FLAGS})
endif()
endif()
@@ -113,7 +112,7 @@ add_dependencies(
# Since USD 21.11 the libraries are prefixed with "usd_", i.e. "libusd_m.a" became "libusd_usd_m.a".
# See https://github.com/PixarAnimationStudios/USD/blob/release/CHANGELOG.md#2111---2021-11-01
if(NOT WIN32)
if(USD_VERSION VERSION_LESS 21.11)
if (USD_VERSION VERSION_LESS 21.11)
set(PXR_LIB_PREFIX "")
else()
set(PXR_LIB_PREFIX "usd_")

View File

@@ -359,7 +359,7 @@ set(SNDFILE_VERSION 1.1.0)
set(SNDFILE_URI https://github.com/libsndfile/libsndfile/releases/download/1.1.0/libsndfile-${SNDFILE_VERSION}.tar.xz)
set(SNDFILE_HASH e63dead2b4f0aaf323687619d007ee6a)
set(SNDFILE_HASH_TYPE MD5)
set(SNDFILE_FILE libsndfile-${SNDFILE_VERSION}.tar.xz)
set(SNDFILE_FILE libsndfile-${SNDFILE_VERSION}.tar.gz)
set(SNDFILE_CPE "cpe:2.3:a:libsndfile_project:libsndfile:${SNDFILE_VERSION}:*:*:*:*:*:*:*")
set(WEBP_VERSION 1.2.2)

View File

@@ -1,4 +1,20 @@
# SPDX-License-Identifier: GPL-2.0-or-later
# ***** BEGIN GPL LICENSE BLOCK *****
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software Foundation,
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ***** END GPL LICENSE BLOCK *****
set(VULKAN_HEADERS_EXTRA_ARGS)
@@ -52,3 +68,4 @@ elseif(WIN32)
)
endif()
endif()

View File

@@ -1,7 +1,5 @@
#!/usr/bin/env python3
# SPDX-License-Identifier: GPL-2.0-or-later
# macOS utility to remove all `rpaths` and add a new one.
# macOS utility to remove all rpaths and add a new one.
import os
import re

View File

@@ -39,15 +39,15 @@ with-all,with-opencollada,with-jack,with-pulseaudio,with-embree,with-oidn,with-n
ver-ocio:,ver-oiio:,ver-llvm:,ver-osl:,ver-osd:,ver-openvdb:,ver-xr-openxr:,ver-level-zero:\
force-all,force-python,force-boost,force-tbb,\
force-ocio,force-imath,force-openexr,force-oiio,force-llvm,force-osl,force-osd,force-openvdb,\
force-ffmpeg,force-opencollada,force-alembic,force-embree,force-oidn,force-materialx,force-usd,\
force-ffmpeg,force-opencollada,force-alembic,force-embree,force-oidn,force-usd,\
force-xr-openxr,force-level-zero,force-openpgl,\
build-all,build-python,build-boost,build-tbb,\
build-ocio,build-imath,build-openexr,build-oiio,build-llvm,build-osl,build-osd,build-openvdb,\
build-ffmpeg,build-opencollada,build-alembic,build-embree,build-oidn,build-materialx,build-usd,\
build-ffmpeg,build-opencollada,build-alembic,build-embree,build-oidn,build-usd,\
build-xr-openxr,build-level-zero,build-openpgl,\
skip-python,skip-boost,skip-tbb,\
skip-ocio,skip-imath,skip-openexr,skip-oiio,skip-llvm,skip-osl,skip-osd,skip-openvdb,\
skip-ffmpeg,skip-opencollada,skip-alembic,skip-embree,skip-oidn,skip-materialx,skip-usd,\
skip-ffmpeg,skip-opencollada,skip-alembic,skip-embree,skip-oidn,skip-usd,\
skip-xr-openxr,skip-level-zero,skip-openpgl \
-- "$@" \
)
@@ -223,9 +223,6 @@ ARGUMENTS_INFO="\"COMMAND LINE ARGUMENTS:
--build-ffmpeg
Force the build of FFMpeg.
--build-materialx
Force the build of MaterialX.
--build-usd
Force the build of Universal Scene Description.
@@ -299,9 +296,6 @@ ARGUMENTS_INFO="\"COMMAND LINE ARGUMENTS:
--force-ffmpeg
Force the rebuild of FFMpeg.
--force-materialx
Force the rebuild of MaterialX.
--force-usd
Force the rebuild of Universal Scene Description.
@@ -368,9 +362,6 @@ ARGUMENTS_INFO="\"COMMAND LINE ARGUMENTS:
--skip-ffmpeg
Unconditionally skip FFMpeg installation/building.
--skip-materialx
Unconditionally skip MaterialX installation/building.
--skip-usd
Unconditionally skip Universal Scene Description installation/building.
@@ -403,7 +394,7 @@ CLANG_FORMAT_VERSION="10.0"
CLANG_FORMAT_VERSION_MIN="6.0"
CLANG_FORMAT_VERSION_MEX="14.0"
PYTHON_VERSION="3.10.9"
PYTHON_VERSION="3.10.8"
PYTHON_VERSION_SHORT="3.10"
PYTHON_VERSION_MIN="3.10"
PYTHON_VERSION_MEX="3.12"
@@ -411,7 +402,6 @@ PYTHON_VERSION_INSTALLED=$PYTHON_VERSION_SHORT
PYTHON_FORCE_BUILD=false
PYTHON_FORCE_REBUILD=false
PYTHON_SKIP=false
_with_built_python=false
# Additional Python modules.
PYTHON_IDNA_VERSION="3.3"
@@ -444,7 +434,7 @@ PYTHON_ZSTANDARD_VERSION_MIN="0.15.2"
PYTHON_ZSTANDARD_VERSION_MEX="0.20.0"
PYTHON_ZSTANDARD_NAME="zstandard"
PYTHON_NUMPY_VERSION="1.23.5"
PYTHON_NUMPY_VERSION="1.22.0"
PYTHON_NUMPY_VERSION_MIN="1.14"
PYTHON_NUMPY_VERSION_MEX="2.0"
PYTHON_NUMPY_NAME="numpy"
@@ -472,13 +462,11 @@ PYTHON_MODULES_PIP=(
)
BOOST_VERSION="1.80.0"
BOOST_VERSION_SHORT="1.80"
BOOST_VERSION="1.78.0"
BOOST_VERSION_SHORT="1.78"
BOOST_VERSION_MIN="1.49"
BOOST_VERSION_MEX="2.0"
# XXX Boost currently has an issue with python/tbb, see rB019b930 for details and patch used to fix it.
# So for now it has to be built, system packages are not usable. :(
BOOST_FORCE_BUILD=true
BOOST_FORCE_BUILD=false
BOOST_FORCE_REBUILD=false
BOOST_SKIP=false
@@ -491,8 +479,8 @@ TBB_FORCE_BUILD=false
TBB_FORCE_REBUILD=false
TBB_SKIP=false
OCIO_VERSION="2.2.0"
OCIO_VERSION_SHORT="2.2"
OCIO_VERSION="2.1.1"
OCIO_VERSION_SHORT="2.1"
OCIO_VERSION_MIN="2.0"
OCIO_VERSION_MEX="3.0"
OCIO_FORCE_BUILD=false
@@ -517,10 +505,10 @@ OPENEXR_FORCE_REBUILD=false
OPENEXR_SKIP=false
_with_built_openexr=false
OIIO_VERSION="2.4.6.0"
OIIO_VERSION_SHORT="2.4"
OIIO_VERSION_MIN="2.2.0"
OIIO_VERSION_MEX="2.5.0"
OIIO_VERSION="2.3.20.0"
OIIO_VERSION_SHORT="2.3"
OIIO_VERSION_MIN="2.1.12"
OIIO_VERSION_MEX="2.4.0"
OIIO_FORCE_BUILD=false
OIIO_FORCE_REBUILD=false
OIIO_SKIP=false
@@ -535,8 +523,8 @@ LLVM_FORCE_REBUILD=false
LLVM_SKIP=false
# OSL needs to be compiled for now!
OSL_VERSION="1.13.0.2"
OSL_VERSION_SHORT="1.13"
OSL_VERSION="1.12.6.2"
OSL_VERSION_SHORT="1.12"
OSL_VERSION_MIN="1.11"
OSL_VERSION_MEX="2.0"
OSL_FORCE_BUILD=false
@@ -544,9 +532,9 @@ OSL_FORCE_REBUILD=false
OSL_SKIP=false
# OpenSubdiv needs to be compiled for now
OSD_VERSION="3.5.0"
OSD_VERSION_SHORT="3.5"
OSD_VERSION_MIN="3.5"
OSD_VERSION="3.4.4"
OSD_VERSION_SHORT="3.4"
OSD_VERSION_MIN="3.4"
OSD_VERSION_MEX="4.0"
OSD_FORCE_BUILD=false
OSD_FORCE_REBUILD=false
@@ -555,10 +543,10 @@ OSD_SKIP=false
# OpenVDB needs to be compiled for now
OPENVDB_BLOSC_VERSION="1.21.1"
OPENVDB_VERSION="10.0.0"
OPENVDB_VERSION_SHORT="10.0"
OPENVDB_VERSION_MIN="10.0"
OPENVDB_VERSION_MEX="11.0"
OPENVDB_VERSION="9.0.0"
OPENVDB_VERSION_SHORT="9.0"
OPENVDB_VERSION_MIN="9.0"
OPENVDB_VERSION_MEX="9.1"
OPENVDB_FORCE_BUILD=false
OPENVDB_FORCE_REBUILD=false
OPENVDB_SKIP=false
@@ -572,16 +560,8 @@ ALEMBIC_FORCE_BUILD=false
ALEMBIC_FORCE_REBUILD=false
ALEMBIC_SKIP=false
MATERIALX_VERSION="1.38.6"
MATERIALX_VERSION_SHORT="1.38"
MATERIALX_VERSION_MIN="1.38"
MATERIALX_VERSION_MEX="1.40"
MATERIALX_FORCE_BUILD=false
MATERIALX_FORCE_REBUILD=false
MATERIALX_SKIP=false
USD_VERSION="22.11"
USD_VERSION_SHORT="22.11"
USD_VERSION="22.03"
USD_VERSION_SHORT="22.03"
USD_VERSION_MIN="20.05"
USD_VERSION_MEX="23.00"
USD_FORCE_BUILD=false
@@ -916,9 +896,6 @@ while true; do
--build-alembic)
ALEMBIC_FORCE_BUILD=true; shift; continue
;;
--build-materialx)
MATERIALX_FORCE_BUILD=true; shift; continue
;;
--build-usd)
USD_FORCE_BUILD=true; shift; continue
;;
@@ -948,7 +925,6 @@ while true; do
OIDN_FORCE_REBUILD=true
FFMPEG_FORCE_REBUILD=true
ALEMBIC_FORCE_REBUILD=true
MATERIALX_FORCE_REBUILD=true
USD_FORCE_REBUILD=true
XR_OPENXR_FORCE_REBUILD=true
LEVEL_ZERO_FORCE_REBUILD=true
@@ -1004,9 +980,6 @@ while true; do
--force-alembic)
ALEMBIC_FORCE_REBUILD=true; shift; continue
;;
--force-materialx)
MATERIALX_FORCE_REBUILD=true; shift; continue
;;
--force-usd)
USD_FORCE_REBUILD=true; shift; continue
;;
@@ -1070,9 +1043,6 @@ while true; do
--skip-usd)
USD_SKIP=true; shift; continue
;;
--skip-materialx)
MATERIALX_SKIP=true; shift; continue
;;
--skip-xr-openxr)
XR_OPENXR_SKIP=true; shift; continue
;;
@@ -1138,9 +1108,7 @@ PYTHON_SOURCE=( "https://www.python.org/ftp/python/$PYTHON_VERSION/Python-$PYTHO
_boost_version_nodots=`echo "$BOOST_VERSION" | sed -r 's/\./_/g'`
BOOST_SOURCE=( "https://boostorg.jfrog.io/artifactory/main/release/$BOOST_VERSION/source/boost_$_boost_version_nodots.tar.bz2" )
BOOST_BUILD_MODULES="--with-filesystem --with-locale --with-thread --with-regex --with-system --with-date_time --with-wave --with-atomic --with-serialization --with-program_options --with-iostreams --with-python"
# Used by debian distros.
BOOST_DEB_PACKAGE_MODULES=( "libboost-filesystem" "libboost-locale" "libboost-thread" "libboost-regex" "libboost-system" "libboost-date-time" "libboost-wave" "libboost-atomic" "libboost-serialization" "libboost-program-options" "libboost-iostreams" "libboost-python" "libboost-numpy" )
BOOST_BUILD_MODULES="--with-system --with-filesystem --with-thread --with-regex --with-locale --with-date_time --with-wave --with-iostreams --with-python --with-program_options --with-serialization --with-atomic"
TBB_SOURCE=( "https://github.com/oneapi-src/oneTBB/archive/$TBB_VERSION$TBB_VERSION_UPDATE.tar.gz" )
TBB_SOURCE_CMAKE=( "https://raw.githubusercontent.com/wjakob/tbb/master/CMakeLists.txt" )
@@ -1170,11 +1138,17 @@ _LLVM_SOURCE_ROOT="https://github.com/llvm/llvm-project/releases/download/llvmor
LLVM_SOURCE=( "$_LLVM_SOURCE_ROOT/llvm-$LLVM_VERSION.src.tar.xz" )
LLVM_CLANG_SOURCE=( "$_LLVM_SOURCE_ROOT/clang-$LLVM_VERSION.src.tar.xz" "$_LLVM_SOURCE_ROOT/cfe-$LLVM_VERSION.src.tar.xz" )
OSL_USE_REPO=true
OSL_USE_REPO=false
OSL_SOURCE=( "https://github.com/imageworks/OpenShadingLanguage/archive/v$OSL_VERSION.tar.gz" )
OSL_SOURCE_REPO=( "https://github.com/AcademySoftwareFoundation/OpenShadingLanguage.git" )
OSL_SOURCE_REPO_BRANCH="main"
OSL_SOURCE_REPO_UID="1a7670600c8b08c2443a78d03c8c27e9a1149140"
#~ OSL_SOURCE_REPO=( "https://github.com/imageworks/OpenShadingLanguage.git" )
#~ OSL_SOURCE_REPO_BRANCH="master"
#~ OSL_SOURCE_REPO_UID="85179714e1bc69cd25ecb6bb711c1a156685d395"
#~ OSL_SOURCE=( "https://github.com/Nazg-Gul/OpenShadingLanguage/archive/Release-1.5.11.tar.gz" )
#~ OSL_SOURCE_REPO=( "https://github.com/mont29/OpenShadingLanguage.git" )
#~ OSL_SOURCE_REPO_UID="85179714e1bc69cd25ecb6bb711c1a156685d395"
#~ OSL_SOURCE_REPO=( "https://github.com/Nazg-Gul/OpenShadingLanguage.git" )
#~ OSL_SOURCE_REPO_UID="7d40ff5fe8e47b030042afb92d0e955f5aa96f48"
#~ OSL_SOURCE_REPO_BRANCH="blender-fixes"
OSD_USE_REPO=false
# Script foo to make the version string compliant with the archive name:
@@ -1197,8 +1171,6 @@ ALEMBIC_SOURCE=( "https://github.com/alembic/alembic/archive/${ALEMBIC_VERSION}.
# ALEMBIC_SOURCE_REPO_UID="e6c90d4faa32c4550adeaaf3f556dad4b73a92bb"
# ALEMBIC_SOURCE_REPO_BRANCH="master"
MATERIALX_SOURCE=( "https://github.com/AcademySoftwareFoundation/MaterialX/archive/refs/tags/v${MATERIALX_VERSION}.tar.gz" )
USD_SOURCE=( "https://github.com/PixarAnimationStudios/USD/archive/v${USD_VERSION}.tar.gz" )
OPENCOLLADA_USE_REPO=false
@@ -1251,10 +1223,8 @@ Those libraries should be available as packages in all recent distributions (opt
* libjpeg, libpng, libtiff, [openjpeg2], [libopenal].
* libx11, libxcursor, libxi, libxrandr, libxinerama (and other libx... as needed).
* libwayland-client0, libdecor, libwayland-cursor0, libwayland-egl1, libxkbcommon0, libdbus-1-3, libegl1 (Wayland)
* libsqlite3, libzstd, libbz2, libssl, libfftw3, libxml2, libtinyxml, yasm, libyaml-cpp, flex, pybind11.
* libsdl2, libepoxy, libpugixml, libpotrace, [libgmp], fontconfig, [libharu/libhpdf].
* [libvulkan/vulkan-loader].
* [libfribidi], [libharfbuzz].\""
* libsqlite3, libzstd, libbz2, libssl, libfftw3, libxml2, libtinyxml, yasm, libyaml-cpp, flex.
* libsdl2, libepoxy, libpugixml, libpotrace, [libgmp], fontconfig, [libharu/libhpdf].\""
DEPS_SPECIFIC_INFO="\"BUILDABLE DEPENDENCIES:
@@ -1515,17 +1485,9 @@ _init_python() {
_update_deps_python() {
if [ "$1" = true ]; then
BOOST_FORCE_BUILD=true
OCIO_FORCE_BUILD=true
OIIO_FORCE_BUILD=true
OPENVDB_FORCE_BUILD=true
USD_FORCE_BUILD=true
fi
if [ "$2" = true ]; then
BOOST_FORCE_REBUILD=true
OCIO_FORCE_REBUILD=true
OIIO_FORCE_REBUILD=true
OPENVDB_FORCE_REBUILD=true
USD_FORCE_REBUILD=true
fi
}
@@ -1611,9 +1573,6 @@ compile_Python() {
PRINT ""
$_python -m pip install $module --no-binary :all:
done
_with_built_python=true
_with_built_python_execpath="$INST/python-$PYTHON_VERSION_SHORT/bin/python3"
}
# ----------------------------------------------------------------------------
@@ -1632,14 +1591,12 @@ _update_deps_boost() {
OSL_FORCE_BUILD=true
OPENVDB_FORCE_BUILD=true
ALEMBIC_FORCE_BUILD=true
USD_FORCE_BUILD=true
fi
if [ "$2" = true ]; then
OIIO_FORCE_REBUILD=true
OSL_FORCE_REBUILD=true
OPENVDB_FORCE_REBUILD=true
ALEMBIC_FORCE_REBUILD=true
USD_FORCE_REBUILD=true
fi
}
@@ -1659,7 +1616,7 @@ compile_Boost() {
fi
# To be changed each time we make edits that would modify the compiled result!
boost_magic=15
boost_magic=14
_init_boost
@@ -1685,13 +1642,11 @@ compile_Boost() {
mkdir -p $SRC
download BOOST_SOURCE[@] $_src.tar.bz2
tar -C $SRC --transform "s,\w*,boost-$BOOST_VERSION,x" -xf $_src.tar.bz2
patch -d $_src -p1 < $SCRIPT_DIR/patches/boost.diff
fi
cd $_src
if [ ! -f $_src/b2 ]; then
if [ -d $_with_built_python ]; then
if [ -d $INST/python-$PYTHON_VERSION_INSTALLED ]; then
./bootstrap.sh --with-python-root="$INST/python-$PYTHON_VERSION_INSTALLED"
else
./bootstrap.sh
@@ -1886,7 +1841,7 @@ compile_OCIO() {
fi
# To be changed each time we make edits that would modify the compiled result!
ocio_magic=5
ocio_magic=3
_init_ocio
# Force having own builds for the dependencies.
@@ -1941,13 +1896,9 @@ compile_OCIO() {
cmake_d="$cmake_d -D CMAKE_PREFIX_PATH=$_inst"
cmake_d="$cmake_d -D CMAKE_INSTALL_PREFIX=$_inst"
cmake_d="$cmake_d -D OCIO_BUILD_APPS=OFF"
cmake_d="$cmake_d -D OCIO_BUILD_PYTHON=ON"
cmake_d="$cmake_d -D OCIO_BUILD_PYTHON=OFF"
cmake_d="$cmake_d -D OCIO_BUILD_GPU_TESTS=OFF"
if [ "$_with_built_python" = true ]; then
cmake_d="$cmake_d -D Python_EXECUTABLE=$_with_built_python_execpath"
fi
if [ $(uname -m) == "aarch64" ]; then
cmake_d="$cmake_d -D OCIO_USE_SSE=OFF"
fi
@@ -2137,13 +2088,11 @@ _update_deps_openexr() {
OIIO_FORCE_BUILD=true
OPENVDB_FORCE_BUILD=true
ALEMBIC_FORCE_BUILD=true
USD_FORCE_BUILD=true
fi
if [ "$2" = true ]; then
OIIO_FORCE_REBUILD=true
OPENVDB_FORCE_REBUILD=true
ALEMBIC_FORCE_REBUILD=true
USD_FORCE_REBUILD=true
fi
}
@@ -2272,11 +2221,9 @@ _init_oiio() {
_update_deps_oiio() {
if [ "$1" = true ]; then
OSL_FORCE_BUILD=true
USD_FORCE_BUILD=true
fi
if [ "$2" = true ]; then
OSL_FORCE_REBUILD=true
USD_FORCE_REBUILD=true
fi
}
@@ -2296,7 +2243,7 @@ compile_OIIO() {
fi
# To be changed each time we make edits that would modify the compiled result!
oiio_magic=20
oiio_magic=18
_init_oiio
# Force having own builds for the dependencies.
@@ -2350,7 +2297,6 @@ compile_OIIO() {
cmake_d="$cmake_d -D CMAKE_INSTALL_PREFIX=$_inst"
cmake_d="$cmake_d -D STOP_ON_WARNING=OFF"
cmake_d="$cmake_d -D LINKSTATIC=OFF"
cmake_d="$cmake_d -D BUILD_SHARED_LIBS=ON"
if [ $(uname -m) != "aarch64" ]; then
cmake_d="$cmake_d -D USE_SIMD=sse2"
@@ -2366,37 +2312,21 @@ compile_OIIO() {
cmake_d="$cmake_d -D OpenEXR_ROOT=$INST/openexr"
fi
cmake_d="$cmake_d -D USE_PYTHON=ON"
if [ "$_with_built_python" = true ]; then
cmake_d="$cmake_d -D Python_EXECUTABLE=$_with_built_python_execpath"
fi
# ptex is only needed when nicholas bishop is ready
cmake_d="$cmake_d -D USE_PTEX=OFF"
# Optional tests and cmd tools
cmake_d="$cmake_d -D USE_QT=OFF"
cmake_d="$cmake_d -D USE_QT5=OFF"
cmake_d="$cmake_d -D USE_OPENGL=OFF"
cmake_d="$cmake_d -D USE_TBB=OFF"
cmake_d="$cmake_d -D USE_BZIP2=OFF"
cmake_d="$cmake_d -D USE_FREETYPE=OFF"
cmake_d="$cmake_d -D USE_OPENCOLORIO=OFF"
cmake_d="$cmake_d -D USE_WEBP=ON"
cmake_d="$cmake_d -D USE_OPENJPEG=ON"
cmake_d="$cmake_d -D USE_PYTHON=OFF"
cmake_d="$cmake_d -D USE_FFMPEG=OFF"
cmake_d="$cmake_d -D USE_OPENCV=OFF"
cmake_d="$cmake_d -D USE_OPENVDB=OFF"
cmake_d="$cmake_d -D USE_NUKE=OFF"
cmake_d="$cmake_d -D USE_DCMTK=OFF"
cmake_d="$cmake_d -D USE_LIBHEIF=OFF"
cmake_d="$cmake_d -D USE_GIF=OFF"
cmake_d="$cmake_d -D USE_LIBRAW=OFF"
cmake_d="$cmake_d -D USE_LIBSQUISH=OFF"
cmake_d="$cmake_d -D BUILD_TESTING=OFF"
cmake_d="$cmake_d -D OIIO_BUILD_TESTS=OFF"
cmake_d="$cmake_d -D OIIO_BUILD_TOOLS=ON"
cmake_d="$cmake_d -D OIIO_BUILD_TOOLS=OFF"
cmake_d="$cmake_d -D TXT2MAN="
#cmake_d="$cmake_d -D CMAKE_EXPORT_COMPILE_COMMANDS=ON"
#cmake_d="$cmake_d -D CMAKE_VERBOSE_MAKEFILE=ON"
if [ -d $INST/boost ]; then
cmake_d="$cmake_d -D BOOST_ROOT=$INST/boost -D Boost_NO_SYSTEM_PATHS=ON -D Boost_NO_BOOST_CMAKE=ON"
@@ -2474,7 +2404,7 @@ compile_LLVM() {
fi
# To be changed each time we make edits that would modify the compiled result!
llvm_magic=4
llvm_magic=3
_init_llvm
# Force having own builds for the dependencies.
@@ -2523,9 +2453,9 @@ compile_LLVM() {
mkdir build
cd build
LLVM_TARGETS="X86;NVPTX"
LLVM_TARGETS="X86"
if [ $(uname -m) == "aarch64" ]; then
LLVM_TARGETS="AArch64;NVPTX"
LLVM_TARGETS="AArch64"
fi
cmake_d="-D CMAKE_BUILD_TYPE=Release"
@@ -2592,7 +2522,7 @@ compile_OSL() {
fi
# To be changed each time we make edits that would modify the compiled result!
osl_magic=22
osl_magic=21
_init_osl
# Force having own builds for the dependencies.
@@ -2623,9 +2553,8 @@ compile_OSL() {
INFO "Unpacking OpenShadingLanguage-$OSL_VERSION"
tar -C $SRC --transform "s,(.*/?)OpenShadingLanguage-[^/]*(.*),\1OpenShadingLanguage-$OSL_VERSION\2,x" \
-xf $_src.tar.gz
patch -d $_src -p1 < $SCRIPT_DIR/patches/osl.diff
fi
patch -d $_src -p1 < $SCRIPT_DIR/patches/osl.diff
fi
cd $_src
@@ -2637,8 +2566,6 @@ compile_OSL() {
# Stick to same rev as windows' libs...
git checkout $OSL_SOURCE_REPO_UID
git reset --hard
patch -d $_src -p1 < $SCRIPT_DIR/patches/osl.diff
fi
# Always refresh the whole build!
@@ -2954,12 +2881,7 @@ _init_openvdb() {
}
_update_deps_openvdb() {
if [ "$1" = true ]; then
USD_FORCE_BUILD=true
fi
if [ "$2" = true ]; then
USD_FORCE_REBUILD=true
fi
:
}
clean_OPENVDB() {
@@ -2981,7 +2903,7 @@ compile_OPENVDB() {
PRINT ""
# To be changed each time we make edits that would modify the compiled result!
openvdb_magic=5
openvdb_magic=4
_init_openvdb
# Force having own builds for the dependencies.
@@ -3030,18 +2952,12 @@ compile_OPENVDB() {
cmake_d="-D CMAKE_BUILD_TYPE=Release"
cmake_d="$cmake_d -D CMAKE_INSTALL_PREFIX=$_inst"
cmake_d="$cmake_d -D USE_STATIC_DEPENDENCIES=OFF"
cmake_d="$cmake_d -D OPENVDB_CORE_SHARED=ON"
cmake_d="$cmake_d -D OPENVDB_CORE_STATIC=OFF"
cmake_d="$cmake_d -D OPENVDB_BUILD_BINARIES=OFF"
if [ "$WITH_NANOVDB" = true ]; then
cmake_d="$cmake_d -D USE_NANOVDB=ON"
cmake_d="$cmake_d -D OPENVDB_BUILD_NANOVDB=ON"
cmake_d="$cmake_d -D NANOVDB_BUILD_TOOLS=OFF"
else
cmake_d="$cmake_d -D USE_NANOVDB=OFF"
cmake_d="$cmake_d -D OPENVDB_BUILD_NANOVDB=OFF"
cmake_d="$cmake_d -D NANOVDB_BUILD_TOOLS=OFF"
fi
if [ -d $INST/boost ]; then
@@ -3053,9 +2969,6 @@ compile_OPENVDB() {
fi
if [ -d $INST/tbb ]; then
cmake_d="$cmake_d -D TBB_ROOT=$INST/tbb"
# Work around until we use oneTBB, otherwise OpenVDB forcefully
# uses oneTBB if it can find it on the system.
cmake_d="$cmake_d -D Tbb_INCLUDE_DIR=$INST/tbb/include"
fi
if [ "$_with_built_imath" = true ]; then
@@ -3069,13 +2982,6 @@ compile_OPENVDB() {
cmake_d="$cmake_d -D Blosc_ROOT=$INST/blosc"
fi
cmake_d="$cmake_d -D OPENVDB_BUILD_PYTHON_MODULE=ON"
cmake_d="$cmake_d -D OPENVDB_PYTHON_WRAP_ALL_GRID_TYPES=ON"
cmake_d="$cmake_d -D USE_NUMPY=ON"
if [ "$_with_built_python" = true ]; then
cmake_d="$cmake_d -D Python_EXECUTABLE=$_with_built_python_execpath"
fi
cmake $cmake_d ..
make -j$THREADS install
@@ -3213,103 +3119,6 @@ compile_ALEMBIC() {
run_ldconfig "alembic"
}
#### Build materialX ####
_init_materialx() {
_src=$SRC/MaterialX-$MATERIALX_VERSION
_git=false
_inst=$INST/materialx-$MATERIALX_VERSION_SHORT
_inst_shortcut=$INST/materialx
}
_update_deps_materialx() {
:
}
clean_MATERIALX() {
_init_materialx
if [ -d $_inst ]; then
# Force rebuilding the dependencies if needed.
_update_deps_materialx false true
fi
_clean
}
compile_MATERIALX() {
if [ "$NO_BUILD" = true ]; then
WARNING "--no-build enabled, MaterialX will not be compiled!"
return
fi
# To be changed each time we make edits that would modify the compiled result!
materialx_magic=1
_init_materialx
# Force having own builds for the dependencies.
_update_deps_materialx true false
# Clean install if needed!
magic_compile_check materialx-$MATERIALX_VERSION $materialx_magic
if [ $? -eq 1 -o "$MATERIALX_FORCE_REBUILD" = true ]; then
clean_MATERIALX
fi
if [ ! -d $_inst ]; then
INFO "Building MaterialX-$MATERIALX_VERSION"
# Force rebuilding the dependencies.
_update_deps_materialx true true
prepare_inst
if [ ! -d $_src ]; then
mkdir -p $SRC
download MATERIALX_SOURCE[@] "$_src.tar.gz"
INFO "Unpacking MaterialX-$MATERIALX_VERSION"
tar -C $SRC -xf $_src.tar.gz
patch -d $_src -p1 < $SCRIPT_DIR/patches/materialx.diff
fi
cd $_src
cmake_d="-D CMAKE_INSTALL_PREFIX=$_inst"
cmake_d="$cmake_d -DMATERIALX_BUILD_SHARED_LIBS=ON"
cmake_d="$cmake_d -DCMAKE_DEBUG_POSTFIX=_d"
cmake_d="$cmake_d -DMATERIALX_BUILD_RENDER=OFF"
cmake_d="$cmake_d -DMATERIALX_BUILD_PYTHON=ON"
cmake_d="$cmake_d -DMATERIALX_INSTALL_PYTHON=OFF"
if [ "$_with_built_python" = true ]; then
cmake_d="$cmake_d -D Python_EXECUTABLE=$_with_built_python_execpath"
fi
cmake $cmake_d ./
make -j$THREADS install
make clean
if [ ! -d $_inst ]; then
ERROR "MaterialX-$MATERIALX_VERSION failed to compile, exiting"
exit 1
fi
magic_compile_set materialx-$MATERIALX_VERSION $materialx_magic
cd $CWD
INFO "Done compiling MaterialX-$MATERIALX_VERSION!"
else
INFO "Own MaterialX-$MATERIALX_VERSION is up to date, nothing to do!"
INFO "If you want to force rebuild of this lib, use the --force-materialx option."
fi
if [ -d $_inst ]; then
_create_inst_shortcut
fi
run_ldconfig "materialx"
}
#### Build USD ####
_init_usd() {
_src=$SRC/USD-$USD_VERSION
@@ -3338,7 +3147,7 @@ compile_USD() {
fi
# To be changed each time we make edits that would modify the compiled result!
usd_magic=2
usd_magic=1
_init_usd
# Force having own builds for the dependencies.
@@ -3372,46 +3181,18 @@ compile_USD() {
cmake_d="-D CMAKE_INSTALL_PREFIX=$_inst"
# For the reasoning behind these options, please see usd.cmake.
if [ -d $INST/boost ]; then
cmake_d="$cmake_d -DBOOST_ROOT=$INST/boost"
cmake_d="$cmake_d $cmake_d -D BOOST_ROOT=$INST/boost"
fi
if [ -d $INST/tbb ]; then
cmake_d="$cmake_d -DTBB_ROOT_DIR=$INST/tbb"
cmake_d="$cmake_d $cmake_d -D TBB_ROOT_DIR=$INST/tbb"
fi
cmake_d="$cmake_d -DPXR_ENABLE_PYTHON_SUPPORT=ON"
cmake_d="$cmake_d -DPXR_USE_PYTHON_3=ON"
if [ "$_with_built_python" = true ]; then
cmake_d="$cmake_d -D PYTHON_EXECUTABLE=$_with_built_python_execpath"
fi
cmake_d="$cmake_d -DPXR_BUILD_IMAGING=ON"
cmake_d="$cmake_d -DPXR_BUILD_OPENIMAGEIO_PLUGIN=ON"
if [ -d $INST/openexr ]; then
cmake_d="$cmake_d -DOPENEXR_LOCATION=$INST/openexr"
fi
if [ -d $INST/oiio ]; then
cmake_d="$cmake_d -DOpenImageIO_ROOT=$INST/oiio"
fi
cmake_d="$cmake_d -DPXR_ENABLE_OPENVDB_SUPPORT=ON"
if [ -d $INST/openvdb ]; then
cmake_d="$cmake_d -DOPENVDB_LOCATION=$INST/openvdb"
fi
cmake_d="$cmake_d -DPXR_ENABLE_GL_SUPPORT=ON"
cmake_d="$cmake_d -DPXR_ENABLE_PYTHON_SUPPORT=OFF"
cmake_d="$cmake_d -DPXR_BUILD_IMAGING=OFF"
cmake_d="$cmake_d -DPXR_BUILD_TESTS=OFF"
cmake_d="$cmake_d -DPXR_BUILD_EXAMPLES=OFF"
cmake_d="$cmake_d -DPXR_BUILD_TUTORIALS=OFF"
cmake_d="$cmake_d -DPXR_BUILD_USD_TOOLS=OFF"
cmake_d="$cmake_d -DPXR_ENABLE_HDF5_SUPPORT=OFF"
cmake_d="$cmake_d -DPXR_ENABLE_MATERIALX_SUPPORT=OFF"
cmake_d="$cmake_d -DPXR_BUILD_USDVIEW=OFF"
cmake_d="$cmake_d -DPXR_BUILD_MONOLITHIC=ON"
cmake_d="$cmake_d -DBUILD_SHARED_LIBS=ON"
cmake_d="$cmake_d -DPXR_BUILD_MONOLITHIC=ON"
cmake_d="$cmake_d -DPXR_BUILD_USD_TOOLS=OFF"
cmake_d="$cmake_d -DCMAKE_DEBUG_POSTFIX=_d"
cmake $cmake_d ./
@@ -4424,12 +4205,11 @@ install_DEB() {
git libfreetype6-dev libfontconfig-dev libx11-dev flex bison libxxf86vm-dev \
libxcursor-dev libxi-dev wget libsqlite3-dev libxrandr-dev libxinerama-dev \
libwayland-dev libdecor-0-dev wayland-protocols libegl-dev libxkbcommon-dev libdbus-1-dev linux-libc-dev \
libvulkan-dev libshaderc-dev \
libbz2-dev libncurses5-dev libssl-dev liblzma-dev libreadline-dev \
libopenal-dev libepoxy-dev yasm pybind11-dev \
libopenal-dev libepoxy-dev yasm \
libsdl2-dev libfftw3-dev patch bzip2 libxml2-dev libtinyxml-dev libjemalloc-dev \
libgmp-dev libpugixml-dev libpotrace-dev libhpdf-dev libzstd-dev libpystring-dev \
libglfw3-dev libfribidi-dev libharfbuzz-dev"
libglfw3-dev"
VORBIS_USE=true
OGG_USE=true
@@ -4613,7 +4393,7 @@ install_DEB() {
boost_version=$(echo `get_package_version_DEB libboost-dev` | sed -r 's/^([0-9]+\.[0-9]+).*/\1/')
install_packages_DEB ${BOOST_DEB_PACKAGE_MODULES[@]/%/$boost_version-dev}
install_packages_DEB libboost-{filesystem,iostreams,locale,regex,system,thread,wave,program-options}$boost_version-dev
clean_Boost
else
compile_Boost
@@ -4805,16 +4585,6 @@ install_DEB() {
compile_ALEMBIC
fi
PRINT ""
if [ "$MATERIALX_SKIP" = true ]; then
WARNING "Skipping MaterialX installation, as requested..."
elif [ "$MATERIALX_FORCE_BUILD" = true ]; then
INFO "Forced MaterialX building, as requested..."
compile_MATERIALX
else
compile_MATERIALX
fi
PRINT ""
if [ "$USD_SKIP" = true ]; then
WARNING "Skipping USD installation, as requested..."
@@ -5158,12 +4928,10 @@ install_RPM() {
libtiff-devel libjpeg-devel libpng-devel sqlite-devel fftw-devel SDL2-devel \
libX11-devel libXi-devel libXcursor-devel libXrandr-devel libXinerama-devel \
wayland-devel libdecor-devel wayland-protocols-devel mesa-libEGL-devel libxkbcommon-devel dbus-devel kernel-headers \
vulkan-loader-devel libshaderc-devel \
wget ncurses-devel readline-devel $OPENJPEG_DEV openal-soft-devel \
libepoxy-devel yasm patch pybind11-devel \
libepoxy-devel yasm patch \
libxml2-devel yaml-cpp-devel tinyxml-devel jemalloc-devel \
gmp-devel pugixml-devel potrace-devel libharu-devel libzstd-devel pystring-devel \
fribidi-devel harfbuzz-devel"
gmp-devel pugixml-devel potrace-devel libharu-devel libzstd-devel pystring-devel"
OPENJPEG_USE=true
VORBIS_USE=true
@@ -5544,16 +5312,6 @@ install_RPM() {
compile_ALEMBIC
fi
PRINT ""
if [ "$MATERIALX_SKIP" = true ]; then
WARNING "Skipping MaterialX installation, as requested..."
elif [ "$MATERIALX_FORCE_BUILD" = true ]; then
INFO "Forced MaterialX building, as requested..."
compile_MATERIALX
else
compile_MATERIALX
fi
PRINT ""
if [ "$USD_SKIP" = true ]; then
WARNING "Skipping USD installation, as requested..."
@@ -5824,10 +5582,9 @@ install_ARCH() {
_packages="$BASE_DEVEL git cmake fontconfig flex \
libxi libxcursor libxrandr libxinerama libepoxy libdecor libpng libtiff wget openal \
vulkan-icd-loader vulkan-headers shaderc \
$OPENJPEG_DEV yasm sdl2 fftw pybind11 \
$OPENJPEG_DEV yasm sdl2 fftw \
libxml2 yaml-cpp tinyxml python-requests jemalloc gmp potrace pugixml libharu \
zstd pystring fribidi harfbuzz"
zstd pystring"
OPENJPEG_USE=true
VORBIS_USE=true
@@ -6159,16 +5916,6 @@ install_ARCH() {
compile_ALEMBIC
fi
PRINT ""
if [ "$MATERIALX_SKIP" = true ]; then
WARNING "Skipping MaterialX installation, as requested..."
elif [ "$MATERIALX_FORCE_BUILD" = true ]; then
INFO "Forced MaterialX building, as requested..."
compile_MATERIALX
else
compile_MATERIALX
fi
PRINT ""
if [ "$USD_SKIP" = true ]; then
WARNING "Skipping USD installation, as requested..."
@@ -6460,27 +6207,6 @@ install_OTHER() {
fi
PRINT ""
if [ "$MATERIALX_SKIP" = true ]; then
WARNING "Skipping MaterialX installation, as requested..."
elif [ "$MATERIALX_FORCE_BUILD" = true ]; then
INFO "Forced MaterialX building, as requested..."
compile_MATERIALX
else
compile_MATERIALX
fi
PRINT ""
if [ "$USD_SKIP" = true ]; then
WARNING "Skipping USD installation, as requested..."
elif [ "$USD_FORCE_BUILD" = true ]; then
INFO "Forced USD building, as requested..."
compile_USD
else
compile_USD
fi
if [ "$WITH_OPENCOLLADA" = true ]; then
PRINT ""
if [ "$OPENCOLLADA_SKIP" = true ]; then
@@ -6558,8 +6284,7 @@ print_info() {
_buildargs="-U *SNDFILE* -U PYTHON* -U *BOOST* -U *Boost* -U *TBB*"
_buildargs="$_buildargs -U *OPENCOLORIO* -U *OPENEXR* -U *OPENIMAGEIO* -U *LLVM* -U *CLANG* -U *CYCLES*"
_buildargs="$_buildargs -U *OPENSUBDIV* -U *OPENVDB* -U *BLOSC* -U *COLLADA* -U *FFMPEG* -U *ALEMBIC*"
_buildargs="$_buildargs -U *MATERIALX* -U *USD*"
_buildargs="$_buildargs -U *OPENSUBDIV* -U *OPENVDB* -U *BLOSC* -U *COLLADA* -U *FFMPEG* -U *ALEMBIC* -U *USD*"
_buildargs="$_buildargs -U *EMBREE* -U *OPENIMAGEDENOISE* -U *OPENXR* -U *OPENPGL*"
_1="-D WITH_CODEC_SNDFILE=ON"
@@ -6746,17 +6471,6 @@ print_info() {
fi
fi
if [ "$MATERIALX_SKIP" = false ]; then
_1="-D WITH_MATERIALX=ON"
PRINT " $_1"
_buildargs="$_buildargs $_1"
if [ -d $INST/materialx ]; then
_1="-D MaterialX_DIR=$INST/materialx/lib/cmake/MaterialX"
PRINT " $_1"
_buildargs="$_buildargs $_1"
fi
fi
if [ "$USD_SKIP" = false ]; then
_1="-D WITH_USD=ON"
PRINT " $_1"

View File

@@ -90,25 +90,3 @@ diff -Naur orig/openvdb/openvdb/tree/ValueAccessor.h openvdb/openvdb/openvdb/tre
CacheItem(TreeCacheT& parent)
: mParent(&parent)
diff --git a/nanovdb/nanovdb/NanoVDB.h b/nanovdb/nanovdb/NanoVDB.h
index f7fc304..fde5c47 100644
--- a/nanovdb/nanovdb/NanoVDB.h
+++ b/nanovdb/nanovdb/NanoVDB.h
@@ -1877,7 +1877,7 @@ __hostdev__ static inline uint32_t FindLowestOn(uint64_t v)
{
NANOVDB_ASSERT(v);
#if (defined(__CUDA_ARCH__) || defined(__HIP__)) && defined(NANOVDB_USE_INTRINSICS)
- return __ffsll(v);
+ return __ffsll(static_cast<unsigned long long int>(v));
#elif defined(_MSC_VER) && defined(NANOVDB_USE_INTRINSICS)
unsigned long index;
_BitScanForward64(&index, v);
@@ -2592,7 +2592,7 @@ public:
///
/// @note This method is only defined for IndexGrid = NanoGrid<ValueIndex>
template <typename T = BuildType>
- __hostdev__ typename enable_if<is_same<T, ValueIndex>::value, uint64_t>::type valueCount() const {return DataType::mData1;}
+ __hostdev__ typename enable_if<is_same<T, ValueIndex>::value, const uint64_t&>::type valueCount() const {return DataType::mData1;}
/// @brief Return a const reference to the tree
__hostdev__ const TreeT& tree() const { return *reinterpret_cast<const TreeT*>(this->treePtr()); }

View File

@@ -27,4 +27,4 @@ set(LIBDIR "${CMAKE_CURRENT_LIST_DIR}/../../../../lib/${LIBDIR_NAME}" CACHE STRI
# Platform specific configuration, to ensure static linking against everything.
# Additional linking libraries
set(CMAKE_EXE_LINKER_FLAGS "-lrt -no-pie" CACHE STRING "" FORCE)
set(CMAKE_EXE_LINKER_FLAGS "-lrt -no-pie" CACHE STRING "" FORCE)

View File

@@ -19,13 +19,9 @@ ENDIF()
SET(_moltenvk_SEARCH_DIRS
${MOLTENVK_ROOT_DIR}
${LIBDIR}/vulkan/MoltenVK
)
# FIXME: These finder modules typically don't use LIBDIR,
# this should be set by `./build_files/cmake/platform/` instead.
IF(DEFINED LIBDIR)
SET(_moltenvk_SEARCH_DIRS ${_moltenvk_SEARCH_DIRS} ${LIBDIR}/vulkan/MoltenVK)
ENDIF()
FIND_PATH(MOLTENVK_INCLUDE_DIR
NAMES

View File

@@ -172,7 +172,7 @@ ENDIF()
# handle the QUIETLY and REQUIRED arguments and set OPENEXR_FOUND to TRUE if
# all listed variables are TRUE
INCLUDE(FindPackageHandleStandardArgs)
FIND_PACKAGE_HANDLE_STANDARD_ARGS(OpenEXR DEFAULT_MSG
FIND_PACKAGE_HANDLE_STANDARD_ARGS(OpenEXR DEFAULT_MSG
_openexr_LIBRARIES OPENEXR_INCLUDE_DIR)
IF(OPENEXR_FOUND)

View File

@@ -17,13 +17,9 @@ ENDIF()
SET(_optix_SEARCH_DIRS
${OPTIX_ROOT_DIR}
"$ENV{PROGRAMDATA}/NVIDIA Corporation/OptiX SDK 7.3.0"
)
# TODO: Which environment uses this?
if(DEFINED ENV{PROGRAMDATA})
list(APPEND _optix_SEARCH_DIRS "$ENV{PROGRAMDATA}/NVIDIA Corporation/OptiX SDK 7.3.0")
endif()
FIND_PATH(OPTIX_INCLUDE_DIR
NAMES
optix.h

View File

@@ -67,8 +67,6 @@ ENDIF()
STRING(REPLACE "." "" PYTHON_VERSION_NO_DOTS ${PYTHON_VERSION})
SET(_PYTHON_ABI_FLAGS "")
SET(_python_SEARCH_DIRS
${PYTHON_ROOT_DIR}
"$ENV{HOME}/py${PYTHON_VERSION_NO_DOTS}"

View File

@@ -6,80 +6,18 @@
import re
import sys
from typing import Optional
cmakelists_file = sys.argv[-1]
def count_backslashes_before_pos(file_data: str, pos: int) -> int:
slash_count = 0
pos -= 1
while pos >= 0:
if file_data[pos] != '\\':
break
pos -= 1
slash_count += 1
return slash_count
def extract_cmake_string_at_pos(file_data: str, pos_beg: int) -> Optional[str]:
assert file_data[pos_beg - 1] == '"'
pos = pos_beg
# Dummy assignment.
pos_end = pos_beg
while True:
pos_next = file_data.find('"', pos)
if pos_next == -1:
raise Exception("Un-terminated string (parse error?)")
count_slashes = count_backslashes_before_pos(file_data, pos_next)
if (count_slashes % 2) == 0:
pos_end = pos_next
# Found the closing quote.
break
# The quote was back-slash escaped, step over it.
pos = pos_next + 1
file_data[pos_next]
assert file_data[pos_end] == '"'
if pos_beg == pos_end:
return None
# See: https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#escape-sequences
text = file_data[pos_beg: pos_end].replace(
# Handle back-slash literals.
"\\\\", "\\",
).replace(
# Handle tabs.
"\\t", "\t",
).replace(
# Handle escaped quotes.
"\\\"", "\"",
).replace(
# Handle tabs.
"\\;", ";",
).replace(
# Handle trailing newlines.
"\\\n", "",
)
return text
def main() -> None:
def main():
options = []
with open(cmakelists_file, 'r', encoding="utf-8") as fh:
file_data = fh.read()
for m in re.finditer(r"^\s*option\s*\(\s*(WITH_[a-zA-Z0-9_]+)\s+(\")", file_data, re.MULTILINE):
option_name = m.group(1)
option_descr = extract_cmake_string_at_pos(file_data, m.span(2)[1])
if option_descr is None:
# Possibly a parsing error, at least show something.
option_descr = "(UNDOCUMENTED)"
options.append("{:s}: {:s}".format(option_name, option_descr))
for l in open(cmakelists_file, 'r').readlines():
if not l.lstrip().startswith('#'):
l_option = re.sub(r'.*\boption\s*\(\s*(WITH_[a-zA-Z0-9_]+)\s+\"(.*)\"\s*.*', r'\g<1> - \g<2>', l)
if l_option != l:
l_option = l_option.strip()
if l_option.startswith('WITH_'):
options.append(l_option)
print('\n'.join(options))

View File

@@ -22,7 +22,7 @@ endif()
# Used for: `source/blender/blenlib/intern/system.c`.
# `execinfo` is not available on non-GLIBC systems (at least not on MUSL-LIBC),
# so check the presence of the header before including it and using the it for back-trace.
# so check the presence of the header before including it and using the it for back-trace.
set(HAVE_EXECINFO_H OFF)
if(NOT MSVC)
include(CheckIncludeFiles)

View File

@@ -326,8 +326,8 @@ function(blender_add_lib__impl
# NOTE: If separated libraries for debug and release are needed every library is the list are
# to be prefixed explicitly.
#
# Use: "optimized libfoo optimized libbar debug libfoo_d debug libbar_d"
# NOT: "optimized libfoo libbar debug libfoo_d libbar_d"
# Use: "optimized libfoo optimized libbar debug libfoo_d debug libbar_d"
# NOT: "optimized libfoo libbar debug libfoo_d libbar_d"
if(NOT "${library_deps}" STREQUAL "")
set(next_library_mode "")
foreach(library ${library_deps})
@@ -535,7 +535,7 @@ function(setup_platform_linker_flags
set_property(TARGET ${target} APPEND_STRING PROPERTY LINK_FLAGS_DEBUG " ${PLATFORM_LINKFLAGS_DEBUG}")
get_target_property(target_type ${target} TYPE)
if(target_type STREQUAL "EXECUTABLE")
if (target_type STREQUAL "EXECUTABLE")
set_property(TARGET ${target} APPEND_STRING PROPERTY LINK_FLAGS " ${PLATFORM_LINKFLAGS_EXECUTABLE}")
endif()
endfunction()
@@ -550,9 +550,7 @@ function(setup_platform_linker_libs
endif()
if(WIN32 AND NOT UNIX)
if(DEFINED PTHREADS_LIBRARIES)
target_link_libraries(${target} ${PTHREADS_LIBRARIES})
endif()
target_link_libraries(${target} ${PTHREADS_LIBRARIES})
endif()
# target_link_libraries(${target} ${PLATFORM_LINKLIBS} ${CMAKE_DL_LIBS})
@@ -1117,7 +1115,7 @@ function(find_python_package
# endif()
# Not set, so initialize.
else()
string(REPLACE "." ";" _PY_VER_SPLIT "${PYTHON_VERSION}")
string(REPLACE "." ";" _PY_VER_SPLIT "${PYTHON_VERSION}")
list(GET _PY_VER_SPLIT 0 _PY_VER_MAJOR)
# re-cache
@@ -1221,10 +1219,10 @@ macro(openmp_delayload
else()
set(OPENMP_DLL_NAME "vcomp140")
endif()
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_RELEASE " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_DEBUG " /DELAYLOAD:${OPENMP_DLL_NAME}d.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_RELWITHDEBINFO " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_MINSIZEREL " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_RELEASE " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_DEBUG " /DELAYLOAD:${OPENMP_DLL_NAME}d.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_RELWITHDEBINFO " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
set_property(TARGET ${projectname} APPEND_STRING PROPERTY LINK_FLAGS_MINSIZEREL " /DELAYLOAD:${OPENMP_DLL_NAME}.dll delayimp.lib")
endif()
endif()
endmacro()
@@ -1264,7 +1262,7 @@ endmacro()
# Utility to gather and install precompiled shared libraries.
macro(add_bundled_libraries library_dir)
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
set(_library_dir ${LIBDIR}/${library_dir})
if(WIN32)
file(GLOB _all_library_versions ${_library_dir}/*\.dll)
@@ -1277,7 +1275,7 @@ macro(add_bundled_libraries library_dir)
list(APPEND PLATFORM_BUNDLED_LIBRARY_DIRS ${_library_dir})
unset(_all_library_versions)
unset(_library_dir)
endif()
endif()
endmacro()
macro(windows_install_shared_manifest)
@@ -1336,13 +1334,13 @@ macro(windows_generate_shared_manifest)
NAME "blender.shared"
)
install(
FILES ${CMAKE_BINARY_DIR}/Release/blender.shared.manifest
DESTINATION "./blender.shared"
CONFIGURATIONS Release;RelWithDebInfo;MinSizeRel
FILES ${CMAKE_BINARY_DIR}/Release/blender.shared.manifest
DESTINATION "./blender.shared"
CONFIGURATIONS Release;RelWithDebInfo;MinSizeRel
)
install(
FILES ${CMAKE_BINARY_DIR}/Debug/blender.shared.manifest
DESTINATION "./blender.shared"
CONFIGURATIONS Debug
FILES ${CMAKE_BINARY_DIR}/Debug/blender.shared.manifest
DESTINATION "./blender.shared"
CONFIGURATIONS Debug
)
endmacro()

View File

@@ -86,14 +86,16 @@ endif()
if(WITH_USD)
find_package(USD REQUIRED)
add_bundled_libraries(usd/lib)
endif()
add_bundled_libraries(usd/lib)
if(WITH_MATERIALX)
find_package(MaterialX)
set_and_warn_library_found("MaterialX" MaterialX_FOUND WITH_MATERIALX)
if(WITH_MATERIALX)
add_bundled_libraries(materialx/lib)
endif()
endif()
add_bundled_libraries(materialx/lib)
if(WITH_VULKAN_BACKEND)
find_package(MoltenVK REQUIRED)
@@ -115,8 +117,8 @@ endif()
if(WITH_OPENSUBDIV)
find_package(OpenSubdiv)
add_bundled_libraries(opensubdiv/lib)
endif()
add_bundled_libraries(opensubdiv/lib)
if(WITH_CODEC_SNDFILE)
find_package(SndFile)
@@ -154,9 +156,9 @@ list(APPEND FREETYPE_LIBRARIES
if(WITH_IMAGE_OPENEXR)
find_package(OpenEXR)
add_bundled_libraries(openexr/lib)
add_bundled_libraries(imath/lib)
endif()
add_bundled_libraries(openexr/lib)
add_bundled_libraries(imath/lib)
if(WITH_CODEC_FFMPEG)
set(FFMPEG_ROOT_DIR ${LIBDIR}/ffmpeg)
@@ -268,11 +270,12 @@ if(WITH_BOOST)
set(BOOST_INCLUDE_DIR ${Boost_INCLUDE_DIRS})
set(BOOST_DEFINITIONS)
add_bundled_libraries(boost/lib)
mark_as_advanced(Boost_LIBRARIES)
mark_as_advanced(Boost_INCLUDE_DIRS)
unset(_boost_FIND_COMPONENTS)
endif()
add_bundled_libraries(boost/lib)
if(WITH_INTERNATIONAL OR WITH_CODEC_FFMPEG)
string(APPEND PLATFORM_LINKFLAGS " -liconv") # boost_locale and ffmpeg needs it !
@@ -294,13 +297,13 @@ if(WITH_OPENIMAGEIO)
)
set(OPENIMAGEIO_DEFINITIONS "-DOIIO_STATIC_BUILD")
set(OPENIMAGEIO_IDIFF "${LIBDIR}/openimageio/bin/idiff")
add_bundled_libraries(openimageio/lib)
endif()
add_bundled_libraries(openimageio/lib)
if(WITH_OPENCOLORIO)
find_package(OpenColorIO 2.0.0 REQUIRED)
add_bundled_libraries(opencolorio/lib)
endif()
add_bundled_libraries(opencolorio/lib)
if(WITH_OPENVDB)
find_package(OpenVDB)
@@ -311,8 +314,8 @@ if(WITH_OPENVDB)
unset(BLOSC_LIBRARIES CACHE)
endif()
set(OPENVDB_DEFINITIONS)
add_bundled_libraries(openvdb/lib)
endif()
add_bundled_libraries(openvdb/lib)
if(WITH_NANOVDB)
find_package(NanoVDB)
@@ -360,8 +363,8 @@ endif()
if(WITH_TBB)
find_package(TBB REQUIRED)
add_bundled_libraries(tbb/lib)
endif()
add_bundled_libraries(tbb/lib)
if(WITH_POTRACE)
find_package(Potrace REQUIRED)
@@ -379,9 +382,9 @@ if(WITH_OPENMP)
set(OpenMP_LIBRARY_DIR "${LIBDIR}/openmp/lib/")
set(OpenMP_LINKER_FLAGS "-L'${OpenMP_LIBRARY_DIR}' -lomp")
set(OpenMP_LIBRARY "${OpenMP_LIBRARY_DIR}/libomp.dylib")
add_bundled_libraries(openmp/lib)
endif()
endif()
add_bundled_libraries(openmp/lib)
if(WITH_XR_OPENXR)
find_package(XR_OpenXR_SDK REQUIRED)

View File

@@ -155,8 +155,8 @@ if("${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64")
# M1 chips run Big Sur onwards.
set(OSX_MIN_DEPLOYMENT_TARGET 11.00)
else()
# 10.15 is our min. target, if you use higher sdk, weak linking happens
set(OSX_MIN_DEPLOYMENT_TARGET 10.15)
# 10.13 is our min. target, if you use higher sdk, weak linking happens
set(OSX_MIN_DEPLOYMENT_TARGET 10.13)
endif()
set(CMAKE_OSX_DEPLOYMENT_TARGET "${OSX_MIN_DEPLOYMENT_TARGET}" CACHE STRING "" FORCE)

View File

@@ -1,47 +1,17 @@
# SPDX-License-Identifier: GPL-2.0-or-later
# Copyright 2022 Blender Foundation. All rights reserved.
# Auto update existing CMake caches for new libraries.
# Auto update existing CMake caches for new libraries
# Assert that `LIBDIR` is defined.
if(NOT (DEFINED LIBDIR))
message(FATAL_ERROR "Logical error, expected 'LIBDIR' to be defined!")
endif()
# Clear cached variables whose name matches `pattern`.
function(unset_cache_variables pattern)
get_cmake_property(_cache_variables CACHE_VARIABLES)
foreach(_cache_variable ${_cache_variables})
foreach (_cache_variable ${_cache_variables})
if("${_cache_variable}" MATCHES "${pattern}")
unset(${_cache_variable} CACHE)
endif()
endforeach()
endfunction()
# Clear cached variables with values containing `contents`.
function(unset_cached_varables_containting contents msg)
get_cmake_property(_cache_variables CACHE_VARIABLES)
set(_found)
set(_print_msg)
foreach(_cache_variable ${_cache_variables})
# Skip "_" prefixed variables, these are used for internal book-keeping,
# not under user control.
string(FIND "${_cache_variable}" "_" _found)
if(NOT (_found EQUAL 0))
string(FIND "${${_cache_variable}}" "${contents}" _found)
if(NOT (_found EQUAL -1))
if(_found)
unset(${_cache_variable} CACHE)
set(_print_msg ON)
endif()
endif()
endif()
endforeach()
if(_print_msg)
message(STATUS ${msg})
endif()
endfunction()
# Detect update from 3.1 to 3.2 libs.
if(UNIX AND
DEFINED OPENEXR_VERSION AND
@@ -93,13 +63,3 @@ if(UNIX AND
unset_cache_variables("^TBB")
unset_cache_variables("^USD")
endif()
if(UNIX AND (NOT APPLE) AND LIBDIR AND (EXISTS ${LIBDIR}))
# Only search for the path if it's found on the system.
set(_libdir_stale "/lib/linux_centos7_x86_64/")
unset_cached_varables_containting(
"${_libdir_stale}"
"Auto clearing old ${_libdir_stale} paths from CMake configuration"
)
unset(_libdir_stale)
endif()

View File

@@ -4,52 +4,38 @@
# Libraries configuration for any *nix system including Linux and Unix (excluding APPLE).
# Detect precompiled library directory
if(NOT DEFINED LIBDIR)
# Path to a locally compiled libraries.
set(LIBDIR_NAME ${CMAKE_SYSTEM_NAME}_${CMAKE_SYSTEM_PROCESSOR})
string(TOLOWER ${LIBDIR_NAME} LIBDIR_NAME)
set(LIBDIR_NATIVE_ABI ${CMAKE_SOURCE_DIR}/../lib/${LIBDIR_NAME})
if(NOT WITH_LIBS_PRECOMPILED)
unset(LIBDIR)
else()
if(NOT DEFINED LIBDIR)
# Path to a locally compiled libraries.
set(LIBDIR_NAME ${CMAKE_SYSTEM_NAME}_${CMAKE_SYSTEM_PROCESSOR})
string(TOLOWER ${LIBDIR_NAME} LIBDIR_NAME)
set(LIBDIR_NATIVE_ABI ${CMAKE_SOURCE_DIR}/../lib/${LIBDIR_NAME})
# Path to precompiled libraries with known glibc 2.28 ABI.
set(LIBDIR_GLIBC228_ABI ${CMAKE_SOURCE_DIR}/../lib/linux_x86_64_glibc_228)
# Path to precompiled libraries with known glibc 2.28 ABI.
set(LIBDIR_GLIBC228_ABI ${CMAKE_SOURCE_DIR}/../lib/linux_x86_64_glibc_228)
# Choose the best suitable libraries.
if(EXISTS ${LIBDIR_NATIVE_ABI})
set(LIBDIR ${LIBDIR_NATIVE_ABI})
# Choose the best suitable libraries.
if(EXISTS ${LIBDIR_NATIVE_ABI})
set(LIBDIR ${LIBDIR_NATIVE_ABI})
set(WITH_LIBC_MALLOC_HOOK_WORKAROUND True)
elseif(EXISTS ${LIBDIR_GLIBC228_ABI})
set(LIBDIR ${LIBDIR_GLIBC228_ABI})
if(WITH_MEM_JEMALLOC)
# jemalloc provides malloc hooks.
set(WITH_LIBC_MALLOC_HOOK_WORKAROUND False)
else()
set(WITH_LIBC_MALLOC_HOOK_WORKAROUND True)
elseif(EXISTS ${LIBDIR_GLIBC228_ABI})
set(LIBDIR ${LIBDIR_GLIBC228_ABI})
if(WITH_MEM_JEMALLOC)
# jemalloc provides malloc hooks.
set(WITH_LIBC_MALLOC_HOOK_WORKAROUND False)
else()
set(WITH_LIBC_MALLOC_HOOK_WORKAROUND True)
endif()
endif()
# Avoid namespace pollustion.
unset(LIBDIR_NATIVE_ABI)
unset(LIBDIR_GLIBC228_ABI)
endif()
if(NOT (EXISTS ${LIBDIR}))
message(STATUS
"Unable to find LIBDIR: ${LIBDIR}, system libraries may be used "
"(disable WITH_LIBS_PRECOMPILED to suppress this message)."
)
unset(LIBDIR)
endif()
# Avoid namespace pollustion.
unset(LIBDIR_NATIVE_ABI)
unset(LIBDIR_GLIBC228_ABI)
endif()
# Support restoring this value once pre-compiled libraries have been handled.
set(WITH_STATIC_LIBS_INIT ${WITH_STATIC_LIBS})
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
message(STATUS "Using pre-compiled LIBDIR: ${LIBDIR}")
file(GLOB LIB_SUBDIRS ${LIBDIR}/*)
@@ -99,7 +85,7 @@ endmacro()
# These are libraries that may be precompiled. For this we disable searching in
# the system directories so that we don't accidentally use them instead.
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
without_system_libs_begin()
endif()
@@ -128,7 +114,7 @@ endfunction()
if(NOT WITH_SYSTEM_FREETYPE)
# FreeType compiled with Brotli compression for woff2.
find_package_wrapper(Freetype REQUIRED)
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
find_package_wrapper(Brotli REQUIRED)
# NOTE: This is done on WIN32 & APPLE but fails on some Linux systems.
@@ -155,7 +141,7 @@ if(WITH_PYTHON)
if(WITH_PYTHON_MODULE AND NOT WITH_INSTALL_PORTABLE)
# Installing into `site-packages`, warn when installing into `./../lib/`
# which script authors almost certainly don't want.
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
path_is_prefix(LIBDIR PYTHON_SITE_PACKAGES _is_prefix)
if(_is_prefix)
message(WARNING "
@@ -180,9 +166,11 @@ endif()
if(WITH_IMAGE_OPENEXR)
find_package_wrapper(OpenEXR) # our own module
set_and_warn_library_found("OpenEXR" OPENEXR_FOUND WITH_IMAGE_OPENEXR)
if(WITH_IMAGE_OPENEXR)
add_bundled_libraries(openexr/lib)
add_bundled_libraries(imath/lib)
endif()
endif()
add_bundled_libraries(openexr/lib)
add_bundled_libraries(imath/lib)
if(WITH_IMAGE_OPENJPEG)
find_package_wrapper(OpenJPEG)
@@ -231,7 +219,7 @@ if(WITH_CODEC_SNDFILE)
endif()
if(WITH_CODEC_FFMPEG)
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
set(FFMPEG_ROOT_DIR ${LIBDIR}/ffmpeg)
# Override FFMPEG components to also include static library dependencies
# included with precompiled libraries, and to ensure correct link order.
@@ -246,7 +234,7 @@ if(WITH_CODEC_FFMPEG)
vpx
x264
xvidcore)
if((DEFINED LIBDIR) AND (EXISTS ${LIBDIR}/ffmpeg/lib/libaom.a))
if(EXISTS ${LIBDIR}/ffmpeg/lib/libaom.a)
list(APPEND FFMPEG_FIND_COMPONENTS aom)
endif()
elseif(FFMPEG)
@@ -340,8 +328,11 @@ endif()
if(WITH_OPENVDB)
find_package(OpenVDB)
set_and_warn_library_found("OpenVDB" OPENVDB_FOUND WITH_OPENVDB)
if(WITH_OPENVDB)
add_bundled_libraries(openvdb/lib)
endif()
endif()
add_bundled_libraries(openvdb/lib)
if(WITH_NANOVDB)
find_package_wrapper(NanoVDB)
@@ -360,14 +351,18 @@ endif()
if(WITH_USD)
find_package_wrapper(USD)
set_and_warn_library_found("USD" USD_FOUND WITH_USD)
if(WITH_USD)
add_bundled_libraries(usd/lib)
endif()
endif()
add_bundled_libraries(usd/lib)
if(WITH_MATERIALX)
find_package_wrapper(MaterialX)
set_and_warn_library_found("MaterialX" MaterialX_FOUND WITH_MATERIALX)
if(WITH_MATERIALX)
add_bundled_libraries(materialx/lib)
endif()
endif()
add_bundled_libraries(materialx/lib)
if(WITH_BOOST)
# uses in build instructions to override include and library variables
@@ -423,8 +418,9 @@ if(WITH_BOOST)
find_package(IcuLinux)
list(APPEND BOOST_LIBRARIES ${ICU_LIBRARIES})
endif()
add_bundled_libraries(boost/lib)
endif()
add_bundled_libraries(boost/lib)
if(WITH_PUGIXML)
find_package_wrapper(PugiXML)
@@ -444,13 +440,10 @@ if(WITH_OPENIMAGEIO)
${PNG_LIBRARIES}
${JPEG_LIBRARIES}
${ZLIB_LIBRARIES}
${BOOST_LIBRARIES}
)
set(OPENIMAGEIO_DEFINITIONS "")
if(WITH_BOOST)
list(APPEND OPENIMAGEIO_LIBRARIES "${BOOST_LIBRARIES}")
endif()
if(WITH_IMAGE_TIFF)
list(APPEND OPENIMAGEIO_LIBRARIES "${TIFF_LIBRARY}")
endif()
@@ -462,16 +455,21 @@ if(WITH_OPENIMAGEIO)
endif()
set_and_warn_library_found("OPENIMAGEIO" OPENIMAGEIO_FOUND WITH_OPENIMAGEIO)
if(WITH_OPENIMAGEIO)
add_bundled_libraries(openimageio/lib)
endif()
endif()
add_bundled_libraries(openimageio/lib)
if(WITH_OPENCOLORIO)
find_package_wrapper(OpenColorIO 2.0.0)
set(OPENCOLORIO_DEFINITIONS "")
set(OPENCOLORIO_DEFINITIONS)
set_and_warn_library_found("OpenColorIO" OPENCOLORIO_FOUND WITH_OPENCOLORIO)
if(WITH_OPENCOLORIO)
add_bundled_libraries(opencolorio/lib)
endif()
endif()
add_bundled_libraries(opencolorio/lib)
if(WITH_CYCLES AND WITH_CYCLES_EMBREE)
find_package(Embree 3.8.0 REQUIRED)
@@ -483,7 +481,7 @@ if(WITH_OPENIMAGEDENOISE)
endif()
if(WITH_LLVM)
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
set(LLVM_STATIC ON)
endif()
@@ -497,7 +495,7 @@ if(WITH_LLVM)
endif()
# Symbol conflicts with same UTF library used by OpenCollada
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
if(WITH_OPENCOLLADA AND (${LLVM_VERSION} VERSION_LESS "4.0.0"))
list(REMOVE_ITEM OPENCOLLADA_LIBRARIES ${OPENCOLLADA_UTF_LIBRARY})
endif()
@@ -512,14 +510,18 @@ if(WITH_OPENSUBDIV)
set(OPENSUBDIV_LIBPATH) # TODO, remove and reference the absolute path everywhere
set_and_warn_library_found("OpenSubdiv" OPENSUBDIV_FOUND WITH_OPENSUBDIV)
if(WITH_OPENSUBDIV)
add_bundled_libraries(opensubdiv/lib)
endif()
endif()
add_bundled_libraries(opensubdiv/lib)
if(WITH_TBB)
find_package_wrapper(TBB)
set_and_warn_library_found("TBB" TBB_FOUND WITH_TBB)
if(WITH_TBB)
add_bundled_libraries(tbb/lib)
endif()
endif()
add_bundled_libraries(tbb/lib)
if(WITH_XR_OPENXR)
find_package(XR_OpenXR_SDK)
@@ -553,7 +555,7 @@ if(WITH_CYCLES AND WITH_CYCLES_PATH_GUIDING)
endif()
endif()
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
without_system_libs_end()
endif()
@@ -568,14 +570,9 @@ else()
endif()
find_package(Threads REQUIRED)
# `FindThreads` documentation notes that this may be empty
# with the system libraries provide threading functionality.
if(CMAKE_THREAD_LIBS_INIT)
list(APPEND PLATFORM_LINKLIBS ${CMAKE_THREAD_LIBS_INIT})
# used by other platforms
set(PTHREADS_LIBRARIES ${CMAKE_THREAD_LIBS_INIT})
endif()
list(APPEND PLATFORM_LINKLIBS ${CMAKE_THREAD_LIBS_INIT})
# used by other platforms
set(PTHREADS_LIBRARIES ${CMAKE_THREAD_LIBS_INIT})
if(CMAKE_DL_LIBS)
list(APPEND PLATFORM_LINKLIBS ${CMAKE_DL_LIBS})
@@ -597,7 +594,7 @@ add_definitions(-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE64_SOURCE
#
# Keep last, so indirectly linked libraries don't override our own pre-compiled libs.
if(DEFINED LIBDIR)
if(EXISTS ${LIBDIR})
# Clear the prefix path as it causes the `LIBDIR` to override system locations.
unset(CMAKE_PREFIX_PATH)
@@ -653,7 +650,7 @@ if(WITH_GHOST_WAYLAND)
# When dynamically linked WAYLAND is used and `${LIBDIR}/wayland` is present,
# there is no need to search for the libraries as they are not needed for building.
# Only the headers are needed which can reference the known paths.
if((DEFINED LIBDIR) AND (EXISTS "${LIBDIR}/wayland" AND WITH_GHOST_WAYLAND_DYNLOAD))
if(EXISTS "${LIBDIR}/wayland" AND WITH_GHOST_WAYLAND_DYNLOAD)
set(_use_system_wayland OFF)
else()
set(_use_system_wayland ON)
@@ -667,7 +664,8 @@ if(WITH_GHOST_WAYLAND)
pkg_check_modules(wayland-protocols wayland-protocols>=1.15)
pkg_get_variable(WAYLAND_PROTOCOLS_DIR wayland-protocols pkgdatadir)
else()
# Rocky8 packages have too old a version, a newer version exist in the pre-compiled libraries.
# CentOS 7 packages have too old a version, a newer version exist in the
# precompiled libraries.
find_path(WAYLAND_PROTOCOLS_DIR
NAMES unstable/xdg-decoration/xdg-decoration-unstable-v1.xml
PATH_SUFFIXES share/wayland-protocols
@@ -717,7 +715,7 @@ if(WITH_GHOST_WAYLAND)
add_definitions(-DWITH_GHOST_WAYLAND_LIBDECOR)
endif()
if((DEFINED LIBDIR) AND (EXISTS "${LIBDIR}/wayland/bin/wayland-scanner"))
if(EXISTS "${LIBDIR}/wayland/bin/wayland-scanner")
set(WAYLAND_SCANNER "${LIBDIR}/wayland/bin/wayland-scanner")
else()
pkg_get_variable(WAYLAND_SCANNER wayland-scanner wayland_scanner)
@@ -792,7 +790,7 @@ if(WITH_GHOST_X11)
endif()
if(WITH_X11_ALPHA)
find_library(X11_Xrender_LIB Xrender ${X11_LIB_SEARCH_PATH})
find_library(X11_Xrender_LIB Xrender ${X11_LIB_SEARCH_PATH})
mark_as_advanced(X11_Xrender_LIB)
if(NOT X11_Xrender_LIB)
message(FATAL_ERROR "libXrender not found. Disable WITH_X11_ALPHA if you

View File

@@ -140,8 +140,8 @@ if(WITH_WINDOWS_BUNDLE_CRT)
install(FILES ${CMAKE_BINARY_DIR}/blender.crt.manifest DESTINATION ./blender.crt)
set(BUNDLECRT "<dependency><dependentAssembly><assemblyIdentity type=\"win32\" name=\"blender.crt\" version=\"1.0.0.0\" /></dependentAssembly></dependency>")
set(BUNDLECRT "${BUNDLECRT}<dependency><dependentAssembly><assemblyIdentity type=\"win32\" name=\"blender.shared\" version=\"1.0.0.0\" /></dependentAssembly></dependency>")
endif()
set(BUNDLECRT "${BUNDLECRT}<dependency><dependentAssembly><assemblyIdentity type=\"win32\" name=\"blender.shared\" version=\"1.0.0.0\" /></dependentAssembly></dependency>")
configure_file(${CMAKE_SOURCE_DIR}/release/windows/manifest/blender.exe.manifest.in ${CMAKE_CURRENT_BINARY_DIR}/blender.exe.manifest @ONLY)
@@ -302,7 +302,7 @@ endif()
file(GLOB children RELATIVE ${LIBDIR} ${LIBDIR}/*)
foreach(child ${children})
if(IS_DIRECTORY ${LIBDIR}/${child})
list(APPEND CMAKE_PREFIX_PATH ${LIBDIR}/${child})
list(APPEND CMAKE_PREFIX_PATH ${LIBDIR}/${child})
endif()
endforeach()
@@ -555,7 +555,7 @@ if(WITH_BOOST)
set(BOOST_PREFIX "")
# This is file new in 3.4 if it does not exist, assume we are building against 3.3 libs
set(BOOST_34_TRIGGER_FILE ${BOOST_LIBPATH}/${BOOST_PREFIX}boost_python310-${BOOST_DEBUG_POSTFIX}.lib)
if(NOT EXISTS ${BOOST_34_TRIGGER_FILE})
if (NOT EXISTS ${BOOST_34_TRIGGER_FILE})
set(BOOST_DEBUG_POSTFIX "vc142-mt-gd-x64-${BOOST_VERSION}")
set(BOOST_PREFIX "lib")
endif()
@@ -573,7 +573,7 @@ if(WITH_BOOST)
debug ${BOOST_LIBPATH}/${BOOST_PREFIX}boost_thread-${BOOST_DEBUG_POSTFIX}.lib
debug ${BOOST_LIBPATH}/${BOOST_PREFIX}boost_chrono-${BOOST_DEBUG_POSTFIX}.lib
)
if(EXISTS ${BOOST_34_TRIGGER_FILE})
if (EXISTS ${BOOST_34_TRIGGER_FILE})
if(WITH_USD)
set(BOOST_PYTHON_LIBRARIES
debug ${BOOST_LIBPATH}/${BOOST_PREFIX}boost_python310-${BOOST_DEBUG_POSTFIX}.lib
@@ -624,7 +624,7 @@ endif()
if(WITH_LLVM)
set(LLVM_ROOT_DIR ${LIBDIR}/llvm CACHE PATH "Path to the LLVM installation")
set(LLVM_INCLUDE_DIRS ${LLVM_ROOT_DIR}/$<$<CONFIG:Debug>:Debug>/include CACHE PATH "Path to the LLVM include directory")
set(LLVM_INCLUDE_DIRS ${LLVM_ROOT_DIR}/$<$<CONFIG:Debug>:Debug>/include CACHE PATH "Path to the LLVM include directory")
file(GLOB LLVM_LIBRARY_OPTIMIZED ${LLVM_ROOT_DIR}/lib/*.lib)
if(EXISTS ${LLVM_ROOT_DIR}/debug/lib)
@@ -1030,7 +1030,7 @@ if(WITH_CYCLES AND WITH_CYCLES_DEVICE_ONEAPI)
${SYCL_ROOT_DIR}/bin/pi_*.dll
)
list(REMOVE_ITEM _sycl_pi_runtime_libraries_glob "${SYCL_ROOT_DIR}/bin/pi_opencl.dll")
list(APPEND _sycl_runtime_libraries ${_sycl_pi_runtime_libraries_glob})
list (APPEND _sycl_runtime_libraries ${_sycl_pi_runtime_libraries_glob})
unset(_sycl_pi_runtime_libraries_glob)
list(APPEND PLATFORM_BUNDLED_LIBRARIES ${_sycl_runtime_libraries})

View File

@@ -43,17 +43,13 @@ update-code:
branch: trunk
commit_id: HEAD
path: lib/benchmarks
assets:
branch: trunk
commit_id: HEAD
path: lib/assets
#
# Buildbot only configs
#
buildbot:
gcc:
version: '11.0.0'
version: '9.0.0'
cuda10:
version: '10.1.243'
cuda11:

View File

@@ -104,30 +104,17 @@ def svn_update(args: argparse.Namespace, release_version: Optional[str]) -> None
svn_url_tests = svn_url + lib_tests
call(svn_non_interactive + ["checkout", svn_url_tests, lib_tests_dirpath])
lib_assets = "assets"
lib_assets_dirpath = os.path.join(lib_dirpath, lib_assets)
if not os.path.exists(lib_assets_dirpath):
print_stage("Checking out Assets")
if make_utils.command_missing(args.svn_command):
sys.stderr.write("svn not found, can't checkout assets\n")
sys.exit(1)
svn_url_assets = svn_url + lib_assets
call(svn_non_interactive + ["checkout", svn_url_assets, lib_assets_dirpath])
# Update precompiled libraries, assets and tests
# Update precompiled libraries and tests
if not os.path.isdir(lib_dirpath):
print("Library path: %r, not found, skipping" % lib_dirpath)
else:
paths_local_and_remote = []
if os.path.exists(os.path.join(lib_dirpath, ".svn")):
print_stage("Updating Precompiled Libraries, Assets and Tests (one repository)")
print_stage("Updating Precompiled Libraries and Tests (one repository)")
paths_local_and_remote.append((lib_dirpath, svn_url))
else:
print_stage("Updating Precompiled Libraries, Assets and Tests (multiple repositories)")
print_stage("Updating Precompiled Libraries and Tests (multiple repositories)")
# Separate paths checked out.
for dirname in os.listdir(lib_dirpath):
if dirname.startswith("."):

View File

@@ -10,9 +10,8 @@ Only keyword arguments can be used to pass operator properties.
Operators don't have return values as you might expect,
instead they return a set() which is made up of:
``{'RUNNING_MODAL', 'CANCELLED', 'FINISHED', 'PASS_THROUGH'}``.
Common return values are ``{'FINISHED'}`` and ``{'CANCELLED'}``, the latter
meaning that the operator execution was aborted without making any changes or
saving an undo history entry.
Common return values are ``{'FINISHED'}`` and ``{'CANCELLED'}``.
Calling an operator in the wrong context will raise a ``RuntimeError``,
there is a poll() method to avoid this problem.

View File

@@ -13,16 +13,16 @@ Blender stores 4 main arrays to define mesh geometry.
- :class:`Mesh.polygons`: (reference a range of loops)
Each polygon references a slice in the loop array, this way, polygons do not store vertices or corner data such as UVs directly,
Each polygon reference a slice in the loop array, this way, polygons do not store vertices or corner data such as UV's directly,
only a reference to loops that the polygon uses.
:class:`Mesh.loops`, :class:`Mesh.uv_layers` :class:`Mesh.vertex_colors` are all aligned so the same polygon loop
indices can be used to find the UVs and vertex colors as with as the vertices.
indices can be used to find the UV's and vertex colors as with as the vertices.
To compare mesh API options see: :ref:`NGons and Tessellation Faces <info_gotcha_mesh_faces>`
This example script prints the vertices and UVs for each polygon, assumes the active object is a mesh with UVs.
This example script prints the vertices and UV's for each polygon, assumes the active object is a mesh with UVs.
"""
import bpy

View File

@@ -44,7 +44,6 @@ class ModalOperator(bpy.types.Operator):
elif event.type == 'LEFTMOUSE': # Confirm
return {'FINISHED'}
elif event.type in {'RIGHTMOUSE', 'ESC'}: # Cancel
# Revert all changes that have been made
context.object.location.x = self.init_loc_x
return {'CANCELLED'}

View File

@@ -7,16 +7,9 @@ This script shows simple operator which prints a message.
Since the operator only has an :class:`Operator.execute` function it takes no
user input.
The function should return ``{'FINISHED'}`` or ``{'CANCELLED'}``, the latter
meaning that operator execution was aborted without making any changes, and
saving an undo entry isn't neccesary. If an error is detected after some changes
have already been made, use the ``{'FINISHED'}`` return code, or the behavior
of undo will be confusing for the user.
.. note::
Operator subclasses must be registered before accessing them from blender.
"""
import bpy

View File

@@ -2098,8 +2098,6 @@ def write_rst_types_index(basepath):
fw(title_string("Types (bpy.types)", "="))
fw(".. module:: bpy.types\n\n")
fw(".. toctree::\n")
# Only show top-level entries (avoids unreasonably large pages).
fw(" :maxdepth: 1\n")
fw(" :glob:\n\n")
fw(" bpy.types.*\n\n")
@@ -2126,8 +2124,6 @@ def write_rst_ops_index(basepath):
write_example_ref("", fw, "bpy.ops")
fw(".. toctree::\n")
fw(" :caption: Submodules\n")
# Only show top-level entries (avoids unreasonably large pages).
fw(" :maxdepth: 1\n")
fw(" :glob:\n\n")
fw(" bpy.ops.*\n\n")
file.close()

View File

@@ -513,19 +513,17 @@ if(WITH_FFTW)
src/fx/Convolver.cpp
src/fx/ConvolverReader.cpp
src/fx/ConvolverSound.cpp
src/fx/Equalizer.cpp
src/fx/FFTConvolver.cpp
src/fx/HRTF.cpp
src/fx/ImpulseResponse.cpp
src/util/FFTPlan.cpp
)
set(FFTW_HDR
include/fx/BinauralSound.h
include/fx/BinauralSound.h
include/fx/BinauralReader.h
include/fx/Convolver.h
include/fx/ConvolverReader.h
include/fx/ConvolverSound.h
include/fx/Equalizer.h
include/fx/FFTConvolver.h
include/fx/HRTF.h
include/fx/HRTFLoader.h

View File

@@ -54,7 +54,6 @@
#ifdef WITH_CONVOLUTION
#include "fx/BinauralSound.h"
#include "fx/ConvolverSound.h"
#include "fx/Equalizer.h"
#endif
#include <cassert>
@@ -769,14 +768,4 @@ AUD_API AUD_Sound* AUD_Sound_Binaural(AUD_Sound* sound, AUD_HRTF* hrtfs, AUD_Sou
}
}
AUD_API AUD_Sound* AUD_Sound_equalize(AUD_Sound* sound, float *definition, int size, float maxFreqEq, int sizeConversion)
{
assert(sound);
std::shared_ptr<Buffer> buf = std::shared_ptr<Buffer>(new Buffer(sizeof(float)*size));
std::memcpy(buf->getBuffer(), definition, sizeof(float)*size);
AUD_Sound *equalizer=new AUD_Sound(new Equalizer(*sound, buf, size, maxFreqEq, sizeConversion));
return equalizer;
}
#endif

View File

@@ -397,16 +397,6 @@ extern AUD_API AUD_Sound* AUD_Sound_mutable(AUD_Sound* sound);
#ifdef WITH_CONVOLUTION
extern AUD_API AUD_Sound* AUD_Sound_Convolver(AUD_Sound* sound, AUD_ImpulseResponse* filter, AUD_ThreadPool* threadPool);
extern AUD_API AUD_Sound* AUD_Sound_Binaural(AUD_Sound* sound, AUD_HRTF* hrtfs, AUD_Source* source, AUD_ThreadPool* threadPool);
/**
* Creates an Equalizer for the sound
* \param sound The handle of the sound
* \param definition buffer of size*sizeof(float) with the array of equalization values
* \param maxFreqEq Maximum frequency refered by the array
* \param sizeConversion Size of the transformation. Must be 2^number (for example 1024, 2048,...)
* \return A handle to the Equalizer refered to that sound
*/
extern AUD_API AUD_Sound* AUD_Sound_equalize(AUD_Sound* sound, float *definition, int size, float maxFreqEq, int sizeConversion);
#endif
#ifdef __cplusplus

View File

@@ -53,7 +53,6 @@ extern AUD_API AUD_Handle* AUD_pauseAfter(AUD_Handle* handle, double seconds);
* \param buffer The buffer to write to. Must have a size of 3*4*length.
* \param length How many samples to read from the sound.
* \param samples_per_second How many samples to read per second of the sound.
* \param interrupt Must point to a short that equals 0. If it is set to a non-zero value, the method will be interrupted and return 0.
* \return How many samples really have been read. Always <= length.
*/
extern AUD_API int AUD_readSound(AUD_Sound* sound, float* buffer, int length, int samples_per_second, short* interrupt);

View File

@@ -5,12 +5,12 @@ import os
import codecs
import numpy
from setuptools import setup, Extension
from distutils.core import setup, Extension
if len(sys.argv) > 2 and sys.argv[1] == '--build-docs':
import subprocess
from setuptools import Distribution
from setuptools.command.build import build
from distutils.core import Distribution
from distutils.command.build import build
dist = Distribution()
cmd = build(dist)

View File

@@ -1,106 +0,0 @@
/*******************************************************************************
* Copyright 2022 Marcos Perez Gonzalez
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
******************************************************************************/
#pragma once
/**
* @file Equalizer.h
* @ingroup fx
* The Equalizer class.
*/
#include <memory>
#include <vector>
#include "ISound.h"
#include "ImpulseResponse.h"
AUD_NAMESPACE_BEGIN
class Buffer;
class ImpulseResponse;
/**
* This class represents a sound that can be modified depending on a given impulse response.
*/
class AUD_API Equalizer : public ISound
{
private:
/**
* A pointer to the imput sound.
*/
std::shared_ptr<ISound> m_sound;
/**
* Local definition of Equalizer
*/
std::shared_ptr<Buffer> m_bufEQ;
/**
* A pointer to the impulse response.
*/
std::shared_ptr<ImpulseResponse> m_impulseResponse;
/**
* delete copy constructor and operator=
*/
Equalizer(const Equalizer&) = delete;
Equalizer& operator=(const Equalizer&) = delete;
/**
* Create ImpulseResponse from the definition in the Buffer,
* using at the end a minimum phase change
*/
std::shared_ptr<ImpulseResponse> createImpulseResponse();
/**
* Create an Impulse Response with minimum phase distortion using Homomorphic
* The input is an Impulse Response
*/
std::shared_ptr<Buffer> minimumPhaseFilterHomomorphic(std::shared_ptr<Buffer> original, int lOriginal, int lWork);
/**
* Create an Impulse Response with minimum phase distortion using Hilbert
* The input is an Impulse Response
*/
std::shared_ptr<Buffer> minimumPhaseFilterHilbert(std::shared_ptr<Buffer> original, int lOriginal, int lWork);
public:
/**
* Creates a new Equalizer.
* \param sound The sound that will be equalized
*/
Equalizer(std::shared_ptr<ISound> sound, std::shared_ptr<Buffer> bufEQ, int externalSizeEq, float maxFreqEq, int sizeConversion);
virtual ~Equalizer();
virtual std::shared_ptr<IReader> createReader();
/*
* Length of the external equalizer definition. It must be the number of "float" positions of the Buffer
*/
int external_size_eq;
/*
* Length of the internal equalizer definition
*/
int filter_length;
/*
* Maximum frequency used in the equalizer definition
*/
float maxFreqEq;
};
AUD_NAMESPACE_END

View File

@@ -27,7 +27,7 @@
AUD_NAMESPACE_BEGIN
BinauralReader::BinauralReader(std::shared_ptr<IReader> reader, std::shared_ptr<HRTF> hrtfs, std::shared_ptr<Source> source, std::shared_ptr<ThreadPool> threadPool, std::shared_ptr<FFTPlan> plan) :
m_position(0), m_reader(reader), m_hrtfs(hrtfs), m_source(source), m_N(plan->getSize()), m_transition(false), m_transPos(CROSSFADE_SAMPLES*NUM_OUTCHANNELS), m_eosReader(false), m_eosTail(false), m_threadPool(threadPool)
m_reader(reader), m_hrtfs(hrtfs), m_source(source), m_N(plan->getSize()), m_threadPool(threadPool), m_position(0), m_eosReader(false), m_eosTail(false), m_transition(false), m_transPos(CROSSFADE_SAMPLES*NUM_OUTCHANNELS)
{
if(m_hrtfs->isEmpty())
AUD_THROW(StateException, "The provided HRTF object is empty");

View File

@@ -23,7 +23,7 @@
AUD_NAMESPACE_BEGIN
Convolver::Convolver(std::shared_ptr<std::vector<std::shared_ptr<std::vector<std::complex<sample_t>>>>> ir, int irLength, std::shared_ptr<ThreadPool> threadPool, std::shared_ptr<FFTPlan> plan) :
m_N(plan->getSize()), m_M(plan->getSize()/2), m_L(plan->getSize()/2), m_irBuffers(ir), m_numThreads(std::min(threadPool->getNumOfThreads(), static_cast<unsigned int>(m_irBuffers->size() - 1))), m_threadPool(threadPool), m_irLength(irLength), m_tailCounter(0), m_eos(false)
m_N(plan->getSize()), m_M(plan->getSize()/2), m_L(plan->getSize()/2), m_irBuffers(ir), m_irLength(irLength), m_threadPool(threadPool), m_numThreads(std::min(threadPool->getNumOfThreads(), static_cast<unsigned int>(m_irBuffers->size() - 1))), m_tailCounter(0), m_eos(false)
{
m_resetFlag = false;

View File

@@ -24,7 +24,7 @@
AUD_NAMESPACE_BEGIN
ConvolverReader::ConvolverReader(std::shared_ptr<IReader> reader, std::shared_ptr<ImpulseResponse> ir, std::shared_ptr<ThreadPool> threadPool, std::shared_ptr<FFTPlan> plan) :
m_position(0), m_reader(reader), m_ir(ir), m_N(plan->getSize()), m_eosReader(false), m_eosTail(false), m_inChannels(reader->getSpecs().channels), m_irChannels(ir->getSpecs().channels), m_threadPool(threadPool)
m_reader(reader), m_ir(ir), m_N(plan->getSize()), m_eosReader(false), m_eosTail(false), m_inChannels(reader->getSpecs().channels), m_irChannels(ir->getSpecs().channels), m_threadPool(threadPool), m_position(0)
{
m_nChannelThreads = std::min((int)threadPool->getNumOfThreads(), m_inChannels);
m_futures.resize(m_nChannelThreads);

View File

@@ -1,367 +0,0 @@
/*******************************************************************************
* Copyright 2022 Marcos Perez Gonzalez
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
******************************************************************************/
#include "fx/Equalizer.h"
#include <chrono>
#include <cstring>
#include <fstream>
#include <iomanip>
#include <iostream>
#include <memory>
#include <string>
#include "Exception.h"
#include "fx/ConvolverReader.h"
#include "fx/ImpulseResponse.h"
#include "util/Buffer.h"
#include "util/FFTPlan.h"
#include "util/ThreadPool.h"
AUD_NAMESPACE_BEGIN
Equalizer::Equalizer(std::shared_ptr<ISound> sound, std::shared_ptr<Buffer> bufEQ, int externalSizeEq, float maxFreqEq, int sizeConversion) : m_sound(sound), m_bufEQ(bufEQ)
{
this->maxFreqEq = maxFreqEq;
this->external_size_eq = externalSizeEq;
filter_length = sizeConversion;
}
Equalizer::~Equalizer()
{
}
std::shared_ptr<IReader> Equalizer::createReader()
{
std::shared_ptr<FFTPlan> fp = std::shared_ptr<FFTPlan>(new FFTPlan(filter_length));
// 2 threads to start with
return std::shared_ptr<ConvolverReader>(new ConvolverReader(m_sound->createReader(), createImpulseResponse(), std::shared_ptr<ThreadPool>(new ThreadPool(2)), fp));
}
float calculateValueArray(float* data, float minX, float maxX, int length, float posX)
{
if(posX < minX)
return 1.0;
if(posX > maxX)
return data[length - 1];
float interval = (maxX - minX) / (float) length;
int idx = (int) ((posX - minX) / interval);
return data[idx];
}
void complex_prod(float a, float b, float c, float d, float* r, float* imag)
{
float prod1 = a * c;
float prod2 = b * d;
float prod3 = (a + b) * (c + d);
// Real Part
*r = prod1 - prod2;
// Imaginary Part
*imag = prod3 - (prod1 + prod2);
}
/**
* The creation of the ImpuseResponse which will be convoluted with the sound
*
* The implementation is based on scikit-signal
*/
std::shared_ptr<ImpulseResponse> Equalizer::createImpulseResponse()
{
std::shared_ptr<FFTPlan> fp = std::shared_ptr<FFTPlan>(new FFTPlan(filter_length));
fftwf_complex* buffer = (fftwf_complex*) fp->getBuffer();
std::memset(buffer, 0, filter_length * sizeof(fftwf_complex));
std::shared_ptr<IReader> soundReader = m_sound.get()->createReader();
Specs specsSound = soundReader.get()->getSpecs();
int sampleRate = specsSound.rate;
for(unsigned i = 0; i < filter_length / 2; i++)
{
double freq = (((float) i) / (float) filter_length) * (float) sampleRate;
double dbGain = calculateValueArray(m_bufEQ->getBuffer(), 0.0, maxFreqEq, external_size_eq, freq);
// gain = 10^(decibels / 20.0)
// 0 db = 1
// 20 db = 10
// 40 db = 100
float gain = (float) pow(10.0, dbGain / 20.0);
if(i == filter_length / 2 - 1)
{
gain = 0;
}
// IMPORTANT!!!! It is needed for the minimum phase step.
// Without this, the amplitude would be square rooted
//
gain *= gain;
// Calculation of exponential with std.. or "by hand"
/*
std::complex<float> preShift= std::complex<float>(0.0, -(filter_length - 1)
/ 2. * M_PI * freq / ( sampleRate/2)); std::complex<float> shift =
std::exp(preShift);
std::complex<float> cGain = gain * shift;
*/
float imaginary_shift = -(filter_length - 1) / 2. * M_PI * freq / (sampleRate / 2);
float cGain_real = gain * cos(imaginary_shift);
float cGain_imag = gain * sin(imaginary_shift);
int i2 = filter_length - i - 1;
buffer[i][0] = cGain_real; // Real
buffer[i][1] = cGain_imag; // Imag
if(i > 0 && i2 < filter_length)
{
buffer[i2][0] = cGain_real; // Real
buffer[i2][1] = cGain_imag; // Imag
}
}
// In place. From Complex to sample_t
fp->IFFT(buffer);
// Window Hamming
sample_t* pt_sample_t = (sample_t*) buffer;
float half_filter = ((float) filter_length) / 2.0;
for(int i = 0; i < filter_length; i++)
{
// Centered in filter_length/2
float window = 0.54 - 0.46 * cos((2 * M_PI * (float) i) / (float) (filter_length - 1));
pt_sample_t[i] *= window;
}
std::shared_ptr<Buffer> b2 = std::shared_ptr<Buffer>(new Buffer(filter_length * sizeof(sample_t)));
sample_t* buffer_real = (sample_t*) buffer;
sample_t* buffer2 = b2->getBuffer();
float normaliziter = (float) filter_length;
for(int i = 0; i < filter_length; i++)
{
buffer2[i] = (buffer_real[i] / normaliziter);
}
fp->freeBuffer(buffer);
//
// Here b2 is the buffer with a "valid" FIR (remember the squared amplitude
//
std::shared_ptr<Buffer> ir_minimum = minimumPhaseFilterHomomorphic(b2, filter_length, -1);
Specs specsIR;
specsIR.rate = sampleRate;
specsIR.channels = CHANNELS_MONO;
return std::shared_ptr<ImpulseResponse>(new ImpulseResponse(std::shared_ptr<StreamBuffer>(new StreamBuffer(ir_minimum, specsIR)), fp));
}
std::shared_ptr<Buffer> Equalizer::minimumPhaseFilterHomomorphic(std::shared_ptr<Buffer> original, int lOriginal, int lWork)
{
void* b_orig = original->getBuffer();
if(lWork < lOriginal || lWork < 0)
{
lWork = (int) pow(2, ceil(log2((float) (2 * (lOriginal - 1) / 0.01))));
}
std::shared_ptr<FFTPlan> fp = std::shared_ptr<FFTPlan>(new FFTPlan(lWork, 0.1));
fftwf_complex* buffer = (fftwf_complex*) fp->getBuffer();
sample_t* b_work = (sample_t*) buffer;
// Padding with 0
std::memset(b_work, 0, lWork * sizeof(sample_t));
std::memcpy(b_work, b_orig, lOriginal * sizeof(sample_t));
fp->FFT(b_work);
for(int i = 0; i < lWork / 2; i++)
{
buffer[i][0] = fabs(sqrt(buffer[i][0] * buffer[i][0] + buffer[i][1] * buffer[i][1]));
buffer[i][1] = 0.0;
int conjugate = lWork - i - 1;
buffer[conjugate][0] = buffer[i][0];
buffer[conjugate][1] = 0.0;
}
double threshold = pow(10.0, -7);
float logThreshold = (float) log(threshold);
// take 0.25*log(|H|**2) = 0.5*log(|H|)
for(int i = 0; i < lWork; i++)
{
if(buffer[i][0] < threshold)
{
buffer[i][0] = 0.5 * logThreshold;
}
else
{
buffer[i][0] = 0.5 * log(buffer[i][0]);
}
}
fp->IFFT(buffer);
// homomorphic filter
int stop = (lOriginal + 1) / 2;
b_work[0] = b_work[0] / (float) lWork;
for(int i = 1; i < stop; i++)
{
b_work[i] = b_work[i] / (float) lWork * 2.0;
}
for(int i = stop; i < lWork; i++)
{
b_work[i] = 0;
}
fp->FFT(buffer);
// EXP
// e^x = e^ (a+bi)= e^a * e^bi = e^a * (cos b + i sin b)
for(int i = 0; i < lWork / 2; i++)
{
float new_real;
float new_imag;
new_real = exp(buffer[i][0]) * cos(buffer[i][1]);
new_imag = exp(buffer[i][0]) * sin(buffer[i][1]);
buffer[i][0] = new_real;
buffer[i][1] = new_imag;
int conjugate = lWork - i - 1;
buffer[conjugate][0] = new_real;
buffer[conjugate][1] = new_imag;
}
// IFFT
fp->IFFT(buffer);
// Create new clean Buffer with only the result and normalization
int lOut = (lOriginal / 2) + lOriginal % 2;
std::shared_ptr<Buffer> bOut = std::shared_ptr<Buffer>(new Buffer(sizeof(float) * lOut));
float* bbOut = (float*) bOut->getBuffer();
// Copy and normalize
for(int i = 0; i < lOut; i++)
{
bbOut[i] = b_work[i] / (float) lWork;
}
fp->freeBuffer(buffer);
return bOut;
}
std::shared_ptr<Buffer> Equalizer::minimumPhaseFilterHilbert(std::shared_ptr<Buffer> original, int lOriginal, int lWork)
{
void* b_orig = original->getBuffer();
if(lWork < lOriginal || lWork < 0)
{
lWork = (int) pow(2, ceil(log2((float) (2 * (lOriginal - 1) / 0.01))));
}
std::shared_ptr<FFTPlan> fp = std::shared_ptr<FFTPlan>(new FFTPlan(lWork, 0.1));
fftwf_complex* buffer = (fftwf_complex*) fp->getBuffer();
sample_t* b_work = (sample_t*) buffer;
// Padding with 0
std::memset(b_work, 0, lWork * sizeof(sample_t));
std::memcpy(b_work, b_orig, lOriginal * sizeof(sample_t));
fp->FFT(b_work);
float mymax, mymin;
float n_half = (float) (lOriginal >> 1);
for(int i = 0; i < lWork; i++)
{
float w = ((float) i) * 2.0 * M_PI / (float) lWork * n_half;
float f1 = cos(w);
float f2 = sin(w);
float f3, f4;
complex_prod(buffer[i][0], buffer[i][1], f1, f2, &f3, &f4);
buffer[i][0] = f3;
buffer[i][1] = 0.0;
if(i == 0)
{
mymax = f3;
mymin = f3;
}
else
{
if(f3 < mymin)
mymin = f3;
if(f3 > mymax)
mymax = f3;
}
}
float dp = mymax - 1;
float ds = 0 - mymin;
float S = 4.0 / pow(2, (sqrt(1 + dp + ds) + sqrt(1 - dp + ds)));
for(int i = 0; i < lWork; i++)
{
buffer[i][0] = sqrt((buffer[i][0] + ds) * S) + 1.0E-10;
}
fftwf_complex* buffer_tmp = (fftwf_complex*) std::malloc(lWork * sizeof(fftwf_complex));
std::memcpy(buffer_tmp, buffer, lWork * sizeof(fftwf_complex));
//
// Hilbert transform
//
int midpt = lWork >> 1;
for(int i = 0; i < lWork; i++)
buffer[i][0] = log(buffer[i][0]);
fp->IFFT(buffer);
b_work[0] = 0.0;
for(int i = 1; i < midpt; i++)
{
b_work[i] /= (float) lWork;
}
b_work[midpt] = 0.0;
for(int i = midpt + 1; i < lWork; i++)
{
b_work[i] /= (-1.0 * lWork);
}
fp->FFT(b_work);
// Exp
for(int i = 0; i < lWork; i++)
{
float base = exp(buffer[i][0]);
buffer[i][0] = base * cos(buffer[i][1]);
buffer[i][1] = base * sin(buffer[i][1]);
complex_prod(buffer_tmp[i][0], buffer_tmp[i][1], buffer[i][0], buffer[i][1], &(buffer[i][0]), &(buffer[i][1]));
}
std::free(buffer_tmp);
fp->IFFT(buffer);
//
// Copy and normalization
//
int n_out = n_half + lOriginal % 2;
std::shared_ptr<Buffer> b_minimum = std::shared_ptr<Buffer>(new Buffer(n_out * sizeof(sample_t)));
std::memcpy(b_minimum->getBuffer(), buffer, n_out * sizeof(sample_t));
sample_t* b_final = (sample_t*) b_minimum->getBuffer();
for(int i = 0; i < n_out; i++)
{
b_final[i] /= (float) lWork;
}
return b_minimum;
}
AUD_NAMESPACE_END

View File

@@ -22,7 +22,7 @@
AUD_NAMESPACE_BEGIN
FFTConvolver::FFTConvolver(std::shared_ptr<std::vector<std::complex<sample_t>>> ir, std::shared_ptr<FFTPlan> plan) :
m_plan(plan), m_N(plan->getSize()), m_M(plan->getSize()/2), m_L(plan->getSize()/2), m_irBuffer(ir), m_tailPos(0)
m_plan(plan), m_N(plan->getSize()), m_M(plan->getSize()/2), m_L(plan->getSize()/2), m_tailPos(0), m_irBuffer(ir)
{
m_tail = (float*)calloc(m_M - 1, sizeof(float));
m_realBufLen = ((m_N / 2) + 1) * 2;

View File

@@ -75,7 +75,7 @@ void HRTFLoader::loadHRTFs(std::shared_ptr<HRTF> hrtfs, char ear, const std::str
if(ear == 'L')
azim = 360 - azim;
}
catch(...)
catch(std::exception& e)
{
AUD_THROW(FileException, "The HRTF name doesn't follow the naming scheme: " + filename);
}
@@ -86,4 +86,4 @@ void HRTFLoader::loadHRTFs(std::shared_ptr<HRTF> hrtfs, char ear, const std::str
return;
}
AUD_NAMESPACE_END
AUD_NAMESPACE_END

View File

@@ -78,7 +78,7 @@ void HRTFLoader::loadHRTFs(std::shared_ptr<HRTF> hrtfs, char ear, const std::str
if(ear == 'L')
azim = 360 - azim;
}
catch(...)
catch(std::exception& e)
{
AUD_THROW(FileException, "The HRTF name doesn't follow the naming scheme: " + filename);
}
@@ -90,4 +90,4 @@ void HRTFLoader::loadHRTFs(std::shared_ptr<HRTF> hrtfs, char ear, const std::str
return;
}
AUD_NAMESPACE_END
AUD_NAMESPACE_END

View File

@@ -13,12 +13,10 @@ endif()
# Exporting functions from the blender binary gives linker warnings on Apple arm64 systems.
# Silence them here.
if(APPLE)
if("${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64")
if(CMAKE_COMPILER_IS_GNUCXX OR "${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
string(APPEND CMAKE_C_FLAGS " -fvisibility=hidden")
string(APPEND CMAKE_CXX_FLAGS " -fvisibility=hidden")
endif()
if(APPLE AND ("${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64"))
if(CMAKE_COMPILER_IS_GNUCXX OR "${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
string(APPEND CMAKE_C_FLAGS " -fvisibility=hidden")
string(APPEND CMAKE_CXX_FLAGS " -fvisibility=hidden")
endif()
endif()
@@ -263,11 +261,9 @@ set(LIB
blender_add_lib(extern_mantaflow "${SRC}" "${INC}" "${INC_SYS}" "${LIB}")
if(WITH_OPENVDB)
# The VDB libs above are only added to as INTERFACE libs by blender_add_lib,
# meaning extern_mantaflow itself actually does not have a dependency on the
# openvdb libraries, and CMAKE is free to link the vdb libs before
# extern_mantaflow causing linker errors on linux. By explicitly declaring
# a dependency here, cmake will do the right thing.
target_link_libraries(extern_mantaflow PRIVATE ${OPENVDB_LIBRARIES})
endif()
# The VDB libs above are only added to as INTERFACE libs by blender_add_lib,
# meaning extern_mantaflow itself actually does not have a dependency on the
# openvdb libraries, and CMAKE is free to link the vdb libs before
# extern_mantaflow causing linker errors on linux. By explicitly declaring
# a dependency here, cmake will do the right thing.
target_link_libraries(extern_mantaflow PRIVATE ${OPENVDB_LIBRARIES})

View File

@@ -85,11 +85,15 @@ elseif(WIN32 AND MSVC AND NOT CMAKE_CXX_COMPILER_ID MATCHES "Clang")
# there is no /arch:SSE3, but intrinsics are available anyway
if(CMAKE_CL_64)
set(CYCLES_SSE2_KERNEL_FLAGS "${CYCLES_KERNEL_FLAGS}")
set(CYCLES_SSE3_KERNEL_FLAGS "${CYCLES_KERNEL_FLAGS}")
set(CYCLES_SSE41_KERNEL_FLAGS "${CYCLES_KERNEL_FLAGS}")
set(CYCLES_AVX_KERNEL_FLAGS "${CYCLES_AVX_ARCH_FLAGS} ${CYCLES_KERNEL_FLAGS}")
set(CYCLES_AVX2_KERNEL_FLAGS "${CYCLES_AVX2_ARCH_FLAGS} ${CYCLES_KERNEL_FLAGS}")
else()
set(CYCLES_SSE2_KERNEL_FLAGS "/arch:SSE2 ${CYCLES_KERNEL_FLAGS}")
set(CYCLES_SSE3_KERNEL_FLAGS "/arch:SSE2 ${CYCLES_KERNEL_FLAGS}")
set(CYCLES_SSE41_KERNEL_FLAGS "/arch:SSE2 ${CYCLES_KERNEL_FLAGS}")
set(CYCLES_AVX_KERNEL_FLAGS "${CYCLES_AVX_ARCH_FLAGS} ${CYCLES_KERNEL_FLAGS}")
set(CYCLES_AVX2_KERNEL_FLAGS "${CYCLES_AVX2_ARCH_FLAGS} ${CYCLES_KERNEL_FLAGS}")
endif()
@@ -122,7 +126,11 @@ elseif(CMAKE_COMPILER_IS_GNUCC OR (CMAKE_CXX_COMPILER_ID MATCHES "Clang"))
endif()
set(CYCLES_SSE2_KERNEL_FLAGS "${CYCLES_KERNEL_FLAGS} -msse -msse2")
set(CYCLES_SSE41_KERNEL_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS} -msse3 -mssse3 -msse4.1")
set(CYCLES_SSE3_KERNEL_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS} -msse3 -mssse3")
set(CYCLES_SSE41_KERNEL_FLAGS "${CYCLES_SSE3_KERNEL_FLAGS} -msse4.1")
if(CXX_HAS_AVX)
set(CYCLES_AVX_KERNEL_FLAGS "${CYCLES_SSE41_KERNEL_FLAGS} -mavx")
endif()
if(CXX_HAS_AVX2)
set(CYCLES_AVX2_KERNEL_FLAGS "${CYCLES_SSE41_KERNEL_FLAGS} -mavx -mavx2 -mfma -mlzcnt -mbmi -mbmi2 -mf16c")
endif()
@@ -136,8 +144,13 @@ elseif(WIN32 AND CMAKE_CXX_COMPILER_ID MATCHES "Intel")
if(CXX_HAS_SSE)
set(CYCLES_SSE2_KERNEL_FLAGS "/QxSSE2")
set(CYCLES_SSE3_KERNEL_FLAGS "/QxSSSE3")
set(CYCLES_SSE41_KERNEL_FLAGS "/QxSSE4.1")
if(CXX_HAS_AVX)
set(CYCLES_AVX_KERNEL_FLAGS "/arch:AVX")
endif()
if(CXX_HAS_AVX2)
set(CYCLES_AVX2_KERNEL_FLAGS "/QxCORE-AVX2")
endif()
@@ -161,8 +174,13 @@ elseif(CMAKE_CXX_COMPILER_ID MATCHES "Intel")
set(CYCLES_SSE2_KERNEL_FLAGS "-xsse2")
endif()
set(CYCLES_SSE3_KERNEL_FLAGS "-xssse3")
set(CYCLES_SSE41_KERNEL_FLAGS "-xsse4.1")
if(CXX_HAS_AVX)
set(CYCLES_AVX_KERNEL_FLAGS "-xavx")
endif()
if(CXX_HAS_AVX2)
set(CYCLES_AVX2_KERNEL_FLAGS "-xcore-avx2")
endif()
@@ -172,10 +190,15 @@ endif()
if(CXX_HAS_SSE)
add_definitions(
-DWITH_KERNEL_SSE2
-DWITH_KERNEL_SSE3
-DWITH_KERNEL_SSE41
)
endif()
if(CXX_HAS_AVX)
add_definitions(-DWITH_KERNEL_AVX)
endif()
if(CXX_HAS_AVX2)
add_definitions(-DWITH_KERNEL_AVX2)
endif()
@@ -275,7 +298,7 @@ if(WITH_CYCLES_DEVICE_METAL)
add_definitions(-DWITH_METAL)
endif()
if(WITH_CYCLES_DEVICE_ONEAPI)
if (WITH_CYCLES_DEVICE_ONEAPI)
add_definitions(-DWITH_ONEAPI)
endif()

View File

@@ -83,7 +83,7 @@ enum_use_layer_samples = (
enum_sampling_pattern = (
('SOBOL_BURLEY', "Sobol-Burley", "Use on-the-fly computed Owen-scrambled Sobol for random sampling", 0),
('TABULATED_SOBOL', "Tabulated Sobol", "Use pre-computed tables of Owen-scrambled Sobol for random sampling", 1),
('TABULATED_SOBOL', "Tabulated Sobol", "Use precomputed tables of Owen-scrambled Sobol for random sampling", 1),
)
enum_emission_sampling = (
@@ -905,8 +905,7 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
use_fast_gi: BoolProperty(
name="Fast GI Approximation",
description="Approximate diffuse indirect light with background tinted ambient occlusion. "
"This provides fast alternative to full global illumination, for interactive viewport rendering or final renders with reduced quality",
description="Approximate diffuse indirect light with background tinted ambient occlusion. This provides fast alternative to full global illumination, for interactive viewport rendering or final renders with reduced quality",
default=False,
)
@@ -951,7 +950,9 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
return _cycles.debug_flags_update(scene)
debug_use_cpu_avx2: BoolProperty(name="AVX2", default=True)
debug_use_cpu_avx: BoolProperty(name="AVX", default=True)
debug_use_cpu_sse41: BoolProperty(name="SSE41", default=True)
debug_use_cpu_sse3: BoolProperty(name="SSE3", default=True)
debug_use_cpu_sse2: BoolProperty(name="SSE2", default=True)
debug_bvh_layout: EnumProperty(
name="BVH Layout",
@@ -1538,24 +1539,10 @@ class CyclesPreferences(bpy.types.AddonPreferences):
use_metalrt: BoolProperty(
name="MetalRT (Experimental)",
description="MetalRT for ray tracing uses less memory for scenes which use curves extensively, and can give better "
"performance in specific cases. However this support is experimental and some scenes may render incorrectly",
description="MetalRT for ray tracing uses less memory for scenes which use curves extensively, and can give better performance in specific cases. However this support is experimental and some scenes may render incorrectly",
default=False,
)
kernel_optimization_level: EnumProperty(
name="Kernel Optimization",
description="Kernels can be optimized based on scene content. Optimized kernels are requested at the start of a render. "
"If optimized kernels are not available, rendering will proceed using generic kernels until the optimized set "
"is available in the cache. This can result in additional CPU usage for a brief time (tens of seconds)",
default='FULL',
items=(
('OFF', "Off", "Disable kernel optimization. Slowest rendering, no extra background CPU usage"),
('INTERSECT', "Intersection only", "Optimize only intersection kernels. Faster rendering, negligible extra background CPU usage"),
('FULL', "Full", "Optimize all kernels. Fastest rendering, may result in extra background CPU usage"),
),
)
def find_existing_device_entry(self, device):
for device_entry in self.devices:
if device_entry.id == device[2] and device_entry.type == device[1]:
@@ -1724,12 +1711,10 @@ class CyclesPreferences(bpy.types.AddonPreferences):
if compute_device_type == 'METAL':
import platform
# MetalRT only works on Apple Silicon at present, pending argument encoding fixes on AMD
# Kernel specialization is only viable on Apple Silicon at present due to relative compilation speed
if platform.machine() == 'arm64':
col = layout.column()
col.use_property_split = True
col.prop(self, "kernel_optimization_level")
col.prop(self, "use_metalrt")
row = layout.row()
row.use_property_split = True
row.prop(self, "use_metalrt")
def draw(self, context):
self.draw_impl(self.layout, context)

View File

@@ -370,7 +370,7 @@ class CYCLES_RENDER_PT_sampling_advanced(CyclesButtonsPanel, Panel):
layout.separator()
heading = layout.column(align=True, heading="Scrambling Distance")
heading.active = cscene.sampling_pattern == 'TABULATED_SOBOL'
heading.active = cscene.sampling_pattern != 'TABULATED_SOBOL'
heading.prop(cscene, "auto_scrambling_distance", text="Automatic")
heading.prop(cscene, "preview_scrambling_distance", text="Viewport")
heading.prop(cscene, "scrambling_distance", text="Multiplier")
@@ -395,13 +395,9 @@ class CYCLES_RENDER_PT_sampling_lights(CyclesButtonsPanel, Panel):
def draw(self, context):
layout = self.layout
layout.use_property_split = True
layout.use_property_decorate = False
scene = context.scene
cscene = scene.cycles
col = layout.column(align=True)
col.prop(cscene, "use_light_tree")
sub = col.row()
sub.prop(cscene, "light_sampling_threshold", text="Light Threshold")
@@ -2112,7 +2108,9 @@ class CYCLES_RENDER_PT_debug(CyclesDebugButtonsPanel, Panel):
row = col.row(align=True)
row.prop(cscene, "debug_use_cpu_sse2", toggle=True)
row.prop(cscene, "debug_use_cpu_sse3", toggle=True)
row.prop(cscene, "debug_use_cpu_sse41", toggle=True)
row.prop(cscene, "debug_use_cpu_avx", toggle=True)
row.prop(cscene, "debug_use_cpu_avx2", toggle=True)
col.prop(cscene, "debug_bvh_layout", text="BVH")
@@ -2358,6 +2356,7 @@ def draw_pause(self, context):
def get_panels():
exclude_panels = {
'DATA_PT_area',
'DATA_PT_camera_dof',
'DATA_PT_falloff_curve',
'DATA_PT_light',

View File

@@ -30,10 +30,7 @@ int blender_device_threads(BL::Scene &b_scene)
return 0;
}
DeviceInfo blender_device_info(BL::Preferences &b_preferences,
BL::Scene &b_scene,
bool background,
bool preview)
DeviceInfo blender_device_info(BL::Preferences &b_preferences, BL::Scene &b_scene, bool background)
{
PointerRNA cscene = RNA_pointer_get(&b_scene.ptr, "cycles");
@@ -116,18 +113,6 @@ DeviceInfo blender_device_info(BL::Preferences &b_preferences,
device.use_metalrt = true;
}
if (preview) {
/* Disable specialization for preview renders. */
device.kernel_optimization_level = KERNEL_OPTIMIZATION_LEVEL_OFF;
}
else {
device.kernel_optimization_level = (KernelOptimizationLevel)get_enum(
cpreferences,
"kernel_optimization_level",
KERNEL_OPTIMIZATION_NUM_LEVELS,
KERNEL_OPTIMIZATION_LEVEL_FULL);
}
return device;
}

View File

@@ -19,8 +19,7 @@ int blender_device_threads(BL::Scene &b_scene);
/* Convert Blender settings to device specification. */
DeviceInfo blender_device_info(BL::Preferences &b_preferences,
BL::Scene &b_scene,
bool background,
bool preview);
bool background);
CCL_NAMESPACE_END

View File

@@ -1,6 +1,12 @@
/* SPDX-License-Identifier: Apache-2.0
* Copyright 2021-2022 Blender Foundation */
#include "blender/display_driver.h"
#include "device/device.h"
#include "util/log.h"
#include "util/math.h"
#include "GPU_context.h"
#include "GPU_immediate.h"
#include "GPU_shader.h"
@@ -9,12 +15,6 @@
#include "RE_engine.h"
#include "blender/display_driver.h"
#include "device/device.h"
#include "util/log.h"
#include "util/math.h"
CCL_NAMESPACE_BEGIN
/* --------------------------------------------------------------------
@@ -721,6 +721,8 @@ static void draw_tile(const float2 &zoom,
return;
}
GPU_texture_bind(texture.gpu_texture, 0);
/* Trick to keep sharp rendering without jagged edges on all GPUs.
*
* The idea here is to enforce driver to use linear interpolation when the image is not zoomed
@@ -733,14 +735,14 @@ static void draw_tile(const float2 &zoom,
const float zoomed_height = draw_tile.params.size.y * zoom.y;
if (texture.width != draw_tile.params.size.x || texture.height != draw_tile.params.size.y) {
/* Resolution divider is different from 1, force nearest interpolation. */
GPU_texture_bind_ex(texture.gpu_texture, GPU_SAMPLER_DEFAULT, 0, false);
GPU_texture_filter_mode(texture.gpu_texture, false);
}
else if (zoomed_width - draw_tile.params.size.x > 0.5f ||
zoomed_height - draw_tile.params.size.y > 0.5f) {
GPU_texture_bind_ex(texture.gpu_texture, GPU_SAMPLER_DEFAULT, 0, false);
GPU_texture_filter_mode(texture.gpu_texture, false);
}
else {
GPU_texture_bind_ex(texture.gpu_texture, GPU_SAMPLER_FILTER, 0, false);
GPU_texture_filter_mode(texture.gpu_texture, true);
}
/* Draw at the parameters for which the texture has been updated for. This allows to always draw

View File

@@ -679,7 +679,7 @@ static void attr_create_pointiness(Scene *scene, Mesh *mesh, BL::Mesh &b_mesh, b
if (num_verts == 0) {
return;
}
const float(*positions)[3] = static_cast<const float(*)[3]>(b_mesh.vertices[0].ptr.data);
const MVert *verts = static_cast<const MVert *>(b_mesh.vertices[0].ptr.data);
/* STEP 1: Find out duplicated vertices and point duplicates to a single
* original vertex.
@@ -765,8 +765,10 @@ static void attr_create_pointiness(Scene *scene, Mesh *mesh, BL::Mesh &b_mesh, b
continue;
}
visited_edges.insert(v0, v1);
float3 co0 = make_float3(positions[v0][0], positions[v0][1], positions[v0][2]);
float3 co1 = make_float3(positions[v1][0], positions[v1][1], positions[v1][2]);
const MVert &b_vert_0 = verts[v0];
const MVert &b_vert_1 = verts[v1];
float3 co0 = make_float3(b_vert_0.co[0], b_vert_0.co[1], b_vert_0.co[2]);
float3 co1 = make_float3(b_vert_1.co[0], b_vert_1.co[1], b_vert_1.co[2]);
float3 edge = normalize(co1 - co0);
edge_accum[v0] += edge;
edge_accum[v1] += -edge;
@@ -917,7 +919,7 @@ static void create_mesh(Scene *scene,
return;
}
const float(*positions)[3] = static_cast<const float(*)[3]>(b_mesh.vertices[0].ptr.data);
const MVert *verts = static_cast<const MVert *>(b_mesh.vertices[0].ptr.data);
if (!subdivision) {
numtris = numfaces;
@@ -940,7 +942,8 @@ static void create_mesh(Scene *scene,
/* create vertex coordinates and normals */
for (int i = 0; i < numverts; i++) {
mesh->add_vertex(make_float3(positions[i][0], positions[i][1], positions[i][2]));
const MVert &b_vert = verts[i];
mesh->add_vertex(make_float3(b_vert.co[0], b_vert.co[1], b_vert.co[2]));
}
AttributeSet &attributes = (subdivision) ? mesh->subd_attributes : mesh->attributes;
@@ -1249,13 +1252,14 @@ void BlenderSync::sync_mesh_motion(BL::Depsgraph b_depsgraph,
float3 *mP = attr_mP->data_float3() + motion_step * numverts;
float3 *mN = (attr_mN) ? attr_mN->data_float3() + motion_step * numverts : NULL;
const float(*positions)[3] = static_cast<const float(*)[3]>(b_mesh.vertices[0].ptr.data);
const MVert *verts = static_cast<const MVert *>(b_mesh.vertices[0].ptr.data);
/* NOTE: We don't copy more that existing amount of vertices to prevent
* possible memory corruption.
*/
for (int i = 0; i < std::min<size_t>(b_verts_num, numverts); i++) {
mP[i] = make_float3(positions[i][0], positions[i][1], positions[i][2]);
const MVert &b_vert = verts[i];
mP[i] = make_float3(b_vert.co[0], b_vert.co[1], b_vert.co[2]);
}
if (mN) {
const float(*b_vert_normals)[3] = static_cast<const float(*)[3]>(

View File

@@ -63,7 +63,9 @@ static void debug_flags_sync_from_scene(BL::Scene b_scene)
PointerRNA cscene = RNA_pointer_get(&b_scene.ptr, "cycles");
/* Synchronize CPU flags. */
flags.cpu.avx2 = get_boolean(cscene, "debug_use_cpu_avx2");
flags.cpu.avx = get_boolean(cscene, "debug_use_cpu_avx");
flags.cpu.sse41 = get_boolean(cscene, "debug_use_cpu_sse41");
flags.cpu.sse3 = get_boolean(cscene, "debug_use_cpu_sse3");
flags.cpu.sse2 = get_boolean(cscene, "debug_use_cpu_sse2");
flags.cpu.bvh_layout = (BVHLayout)get_enum(cscene, "debug_bvh_layout");
/* Synchronize CUDA flags. */
@@ -752,7 +754,7 @@ static PyObject *denoise_func(PyObject * /*self*/, PyObject *args, PyObject *key
RNA_id_pointer_create((ID *)PyLong_AsVoidPtr(pyscene), &sceneptr);
BL::Scene b_scene(sceneptr);
DeviceInfo device = blender_device_info(b_preferences, b_scene, true, true);
DeviceInfo device = blender_device_info(b_preferences, b_scene, true);
/* Get denoising parameters from view layer. */
PointerRNA viewlayerptr;

View File

@@ -26,7 +26,7 @@
CCL_NAMESPACE_BEGIN
typedef unordered_multimap<void *, ShaderInput *> PtrInputMap;
typedef map<void *, ShaderInput *> PtrInputMap;
typedef map<void *, ShaderOutput *> PtrOutputMap;
typedef map<string, ConvertNode *> ProxyMap;
@@ -1251,9 +1251,7 @@ static void add_nodes(Scene *scene,
ConvertNode *proxy = graph->create_node<ConvertNode>(to_socket_type, to_socket_type, true);
/* Muted nodes can result in multiple Cycles input sockets mapping to the same Blender
* input socket, so this needs to be a multimap. */
input_map.emplace(b_link.from_socket().ptr.data, proxy->inputs[0]);
input_map[b_link.from_socket().ptr.data] = proxy->inputs[0];
output_map[b_link.to_socket().ptr.data] = proxy->outputs[0];
graph->add(proxy);
@@ -1288,7 +1286,7 @@ static void add_nodes(Scene *scene,
/* register the proxy node for internal binding */
group_proxy_input_map[b_input.identifier()] = proxy;
input_map.emplace(b_input.ptr.data, proxy->inputs[0]);
input_map[b_input.ptr.data] = proxy->inputs[0];
set_default_value(proxy->inputs[0], b_input, b_data, b_ntree);
}
@@ -1340,7 +1338,7 @@ static void add_nodes(Scene *scene,
if (proxy_it != proxy_output_map.end()) {
ConvertNode *proxy = proxy_it->second;
input_map.emplace(b_input.ptr.data, proxy->inputs[0]);
input_map[b_input.ptr.data] = proxy->inputs[0];
set_default_value(proxy->inputs[0], b_input, b_data, b_ntree);
}
@@ -1371,7 +1369,7 @@ static void add_nodes(Scene *scene,
/* XXX should not happen, report error? */
continue;
}
input_map.emplace(b_input.ptr.data, input);
input_map[b_input.ptr.data] = input;
set_default_value(input, b_input, b_data, b_ntree);
}
@@ -1403,23 +1401,20 @@ static void add_nodes(Scene *scene,
BL::NodeSocket b_from_sock = b_link.from_socket();
BL::NodeSocket b_to_sock = b_link.to_socket();
ShaderOutput *output = nullptr;
ShaderOutput *output = 0;
ShaderInput *input = 0;
PtrOutputMap::iterator output_it = output_map.find(b_from_sock.ptr.data);
if (output_it != output_map.end())
output = output_it->second;
PtrInputMap::iterator input_it = input_map.find(b_to_sock.ptr.data);
if (input_it != input_map.end())
input = input_it->second;
/* either socket may be NULL when the node was not exported, typically
/* either node may be NULL when the node was not exported, typically
* because the node type is not supported */
if (output != nullptr) {
ShaderOutput *output = output_it->second;
auto inputs = input_map.equal_range(b_to_sock.ptr.data);
for (PtrInputMap::iterator input_it = inputs.first; input_it != inputs.second; ++input_it) {
ShaderInput *input = input_it->second;
if (input != nullptr) {
graph->connect(output, input);
}
}
}
if (output && input)
graph->connect(output, input);
}
}

View File

@@ -866,8 +866,7 @@ SessionParams BlenderSync::get_session_params(BL::RenderEngine &b_engine,
/* Device */
params.threads = blender_device_threads(b_scene);
params.device = blender_device_info(
b_preferences, b_scene, params.background, b_engine.is_preview());
params.device = blender_device_info(b_preferences, b_scene, params.background);
/* samples */
int samples = get_int(cscene, "samples");

View File

@@ -45,8 +45,8 @@ macro(cycles_add_library target library_deps)
# NOTE: If separated libraries for debug and release ar eneeded every library is the list are
# to be prefixed explicitly.
#
# Use: "optimized libfoo optimized libbar debug libfoo_d debug libbar_d"
# NOT: "optimized libfoo libbar debug libfoo_d libbar_d"
# Use: "optimized libfoo optimized libbar debug libfoo_d debug libbar_d"
# NOT: "optimized libfoo libbar debug libfoo_d libbar_d"
#
# TODO(sergey): This is the same as Blender's side CMake. Find a way to avoid duplication
# somehow in a way which allows to have Cycles standalone.
@@ -111,10 +111,8 @@ macro(cycles_external_libraries_append libraries)
endif()
if(WITH_OPENIMAGEDENOISE)
list(APPEND ${libraries} ${OPENIMAGEDENOISE_LIBRARIES})
if(APPLE)
if("${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64")
list(APPEND ${libraries} "-framework Accelerate")
endif()
if(APPLE AND "${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64")
list(APPEND ${libraries} "-framework Accelerate")
endif()
endif()
if(WITH_ALEMBIC)
@@ -138,15 +136,7 @@ macro(cycles_external_libraries_append libraries)
${PYTHON_LIBRARIES}
${ZLIB_LIBRARIES}
${CMAKE_DL_LIBS}
)
if(DEFINED PTHREADS_LIBRARIES)
list(APPEND ${libraries}
${PTHREADS_LIBRARIES}
)
endif()
list(APPEND ${libraries}
${PTHREADS_LIBRARIES}
${PLATFORM_LINKLIBS}
)

View File

@@ -162,10 +162,7 @@ if(WITH_CYCLES_DEVICE_METAL)
)
endif()
if(WITH_CYCLES_DEVICE_ONEAPI)
if(WITH_CYCLES_ONEAPI_HOST_TASK_EXECUTION)
add_definitions(-DWITH_ONEAPI_SYCL_HOST_TASK)
endif()
if (WITH_CYCLES_DEVICE_ONEAPI)
if(WITH_CYCLES_ONEAPI_BINARIES)
set(cycles_kernel_oneapi_lib_suffix "_aot")
else()

View File

@@ -45,7 +45,9 @@ string device_cpu_capabilities()
{
string capabilities = "";
capabilities += system_cpu_support_sse2() ? "SSE2 " : "";
capabilities += system_cpu_support_sse3() ? "SSE3 " : "";
capabilities += system_cpu_support_sse41() ? "SSE41 " : "";
capabilities += system_cpu_support_avx() ? "AVX " : "";
capabilities += system_cpu_support_avx2() ? "AVX2" : "";
if (capabilities[capabilities.size() - 1] == ' ')
capabilities.resize(capabilities.size() - 1);

View File

@@ -9,7 +9,8 @@ CCL_NAMESPACE_BEGIN
#define KERNEL_FUNCTIONS(name) \
KERNEL_NAME_EVAL(cpu, name), KERNEL_NAME_EVAL(cpu_sse2, name), \
KERNEL_NAME_EVAL(cpu_sse41, name), KERNEL_NAME_EVAL(cpu_avx2, name)
KERNEL_NAME_EVAL(cpu_sse3, name), KERNEL_NAME_EVAL(cpu_sse41, name), \
KERNEL_NAME_EVAL(cpu_avx, name), KERNEL_NAME_EVAL(cpu_avx2, name)
#define REGISTER_KERNEL(name) name(KERNEL_FUNCTIONS(name))
#define REGISTER_KERNEL_FILM_CONVERT(name) \

View File

@@ -17,10 +17,13 @@ template<typename FunctionType> class CPUKernelFunction {
public:
CPUKernelFunction(FunctionType kernel_default,
FunctionType kernel_sse2,
FunctionType kernel_sse3,
FunctionType kernel_sse41,
FunctionType kernel_avx,
FunctionType kernel_avx2)
{
kernel_info_ = get_best_kernel_info(kernel_default, kernel_sse2, kernel_sse41, kernel_avx2);
kernel_info_ = get_best_kernel_info(
kernel_default, kernel_sse2, kernel_sse3, kernel_sse41, kernel_avx, kernel_avx2);
}
template<typename... Args> inline auto operator()(Args... args) const
@@ -57,12 +60,16 @@ template<typename FunctionType> class CPUKernelFunction {
KernelInfo get_best_kernel_info(FunctionType kernel_default,
FunctionType kernel_sse2,
FunctionType kernel_sse3,
FunctionType kernel_sse41,
FunctionType kernel_avx,
FunctionType kernel_avx2)
{
/* Silence warnings about unused variables when compiling without some architectures. */
(void)kernel_sse2;
(void)kernel_sse3;
(void)kernel_sse41;
(void)kernel_avx;
(void)kernel_avx2;
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2
@@ -71,12 +78,24 @@ template<typename FunctionType> class CPUKernelFunction {
}
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX
if (DebugFlags().cpu.has_avx() && system_cpu_support_avx()) {
return KernelInfo("AVX", kernel_avx);
}
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41
if (DebugFlags().cpu.has_sse41() && system_cpu_support_sse41()) {
return KernelInfo("SSE4.1", kernel_sse41);
}
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3
if (DebugFlags().cpu.has_sse3() && system_cpu_support_sse3()) {
return KernelInfo("SSE3", kernel_sse3);
}
#endif
#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2
if (DebugFlags().cpu.has_sse2() && system_cpu_support_sse2()) {
return KernelInfo("SSE2", kernel_sse2);

View File

@@ -57,32 +57,22 @@ enum DeviceTypeMask {
#define DEVICE_MASK(type) (DeviceTypeMask)(1 << type)
enum KernelOptimizationLevel {
KERNEL_OPTIMIZATION_LEVEL_OFF = 0,
KERNEL_OPTIMIZATION_LEVEL_INTERSECT = 1,
KERNEL_OPTIMIZATION_LEVEL_FULL = 2,
KERNEL_OPTIMIZATION_NUM_LEVELS
};
class DeviceInfo {
public:
DeviceType type;
string description;
string id; /* used for user preferences, should stay fixed with changing hardware config */
int num;
bool display_device; /* GPU is used as a display device. */
bool has_nanovdb; /* Support NanoVDB volumes. */
bool has_light_tree; /* Support light tree. */
bool has_osl; /* Support Open Shading Language. */
bool has_guiding; /* Support path guiding. */
bool has_profiling; /* Supports runtime collection of profiling info. */
bool has_peer_memory; /* GPU has P2P access to memory of another GPU. */
bool has_gpu_queue; /* Device supports GPU queue. */
bool use_metalrt; /* Use MetalRT to accelerate ray queries (Metal only). */
KernelOptimizationLevel kernel_optimization_level; /* Optimization level applied to path tracing
kernels (Metal only). */
DenoiserTypeMask denoisers; /* Supported denoiser types. */
bool display_device; /* GPU is used as a display device. */
bool has_nanovdb; /* Support NanoVDB volumes. */
bool has_light_tree; /* Support light tree. */
bool has_osl; /* Support Open Shading Language. */
bool has_guiding; /* Support path guiding. */
bool has_profiling; /* Supports runtime collection of profiling info. */
bool has_peer_memory; /* GPU has P2P access to memory of another GPU. */
bool has_gpu_queue; /* Device supports GPU queue. */
bool use_metalrt; /* Use MetalRT to accelerate ray queries (Metal only). */
DenoiserTypeMask denoisers; /* Supported denoiser types. */
int cpu_threads;
vector<DeviceInfo> multi_devices;
string error_msg;
@@ -177,17 +167,6 @@ class Device {
return true;
}
/* Request cancellation of any long-running work. */
virtual void cancel()
{
}
/* Return true if device is ready for rendering, or report status if not. */
virtual bool is_ready(string & /*status*/) const
{
return true;
}
/* GPU device only functions.
* These may not be used on CPU or multi-devices. */

View File

@@ -76,20 +76,7 @@ class MetalDevice : public Device {
bool use_metalrt = false;
MetalPipelineType kernel_specialization_level = PSO_GENERIC;
int device_id = 0;
static thread_mutex existing_devices_mutex;
static std::map<int, MetalDevice *> active_device_ids;
static bool is_device_cancelled(int device_id);
static MetalDevice *get_device_by_ID(int device_idID,
thread_scoped_lock &existing_devices_mutex_lock);
virtual bool is_ready(string &status) const override;
virtual void cancel() override;
std::atomic_bool async_compile_and_load = false;
virtual BVHLayoutMask get_bvh_layout_mask() const override;
@@ -105,12 +92,14 @@ class MetalDevice : public Device {
bool use_adaptive_compilation();
bool make_source_and_check_if_compile_needed(MetalPipelineType pso_type);
void make_source(MetalPipelineType pso_type, const uint kernel_features);
virtual bool load_kernels(const uint kernel_features) override;
void reserve_local_memory(const uint kernel_features);
void init_host_memory();
void load_texture_info();
void erase_allocation(device_memory &mem);
@@ -123,7 +112,7 @@ class MetalDevice : public Device {
virtual void optimize_for_scene(Scene *scene) override;
static void compile_and_load(int device_id, MetalPipelineType pso_type);
bool compile_and_load(MetalPipelineType pso_type);
/* ------------------------------------------------------------------ */
/* low-level memory management */

View File

@@ -13,32 +13,10 @@
# include "util/path.h"
# include "util/time.h"
# include <crt_externs.h>
CCL_NAMESPACE_BEGIN
class MetalDevice;
thread_mutex MetalDevice::existing_devices_mutex;
std::map<int, MetalDevice *> MetalDevice::active_device_ids;
/* Thread-safe device access for async work. Calling code must pass an appropriately scoped lock
* to existing_devices_mutex to safeguard against destruction of the returned instance. */
MetalDevice *MetalDevice::get_device_by_ID(int ID, thread_scoped_lock &existing_devices_mutex_lock)
{
auto it = active_device_ids.find(ID);
if (it != active_device_ids.end()) {
return it->second;
}
return nullptr;
}
bool MetalDevice::is_device_cancelled(int ID)
{
thread_scoped_lock lock(existing_devices_mutex);
return get_device_by_ID(ID, lock) == nullptr;
}
BVHLayoutMask MetalDevice::get_bvh_layout_mask() const
{
return use_metalrt ? BVH_LAYOUT_METAL : BVH_LAYOUT_BVH2;
@@ -62,15 +40,6 @@ void MetalDevice::set_error(const string &error)
MetalDevice::MetalDevice(const DeviceInfo &info, Stats &stats, Profiler &profiler)
: Device(info, stats, profiler), texture_info(this, "texture_info", MEM_GLOBAL)
{
{
/* Assign an ID for this device which we can use to query whether async shader compilation
* requests are still relevant. */
thread_scoped_lock lock(existing_devices_mutex);
static int existing_devices_counter = 1;
device_id = existing_devices_counter++;
active_device_ids[device_id] = this;
}
mtlDevId = info.num;
/* select chosen device */
@@ -88,6 +57,7 @@ MetalDevice::MetalDevice(const DeviceInfo &info, Stats &stats, Profiler &profile
if (@available(macos 11.0, *)) {
if ([mtlDevice hasUnifiedMemory]) {
default_storage_mode = MTLResourceStorageModeShared;
init_host_memory();
}
}
@@ -110,6 +80,10 @@ MetalDevice::MetalDevice(const DeviceInfo &info, Stats &stats, Profiler &profile
case METAL_GPU_APPLE: {
max_threads_per_threadgroup = 512;
use_metalrt = info.use_metalrt;
/* Specialize the intersection kernels on Apple GPUs by default as these can be built very
* quickly. */
kernel_specialization_level = PSO_SPECIALIZED_INTERSECT;
break;
}
}
@@ -122,22 +96,6 @@ MetalDevice::MetalDevice(const DeviceInfo &info, Stats &stats, Profiler &profile
capture_enabled = true;
}
if (device_vendor == METAL_GPU_APPLE) {
/* Set kernel_specialization_level based on user prefs. */
switch (info.kernel_optimization_level) {
case KERNEL_OPTIMIZATION_LEVEL_OFF:
kernel_specialization_level = PSO_GENERIC;
break;
default:
case KERNEL_OPTIMIZATION_LEVEL_INTERSECT:
kernel_specialization_level = PSO_SPECIALIZED_INTERSECT;
break;
case KERNEL_OPTIMIZATION_LEVEL_FULL:
kernel_specialization_level = PSO_SPECIALIZED_SHADE;
break;
}
}
if (auto envstr = getenv("CYCLES_METAL_SPECIALIZATION_LEVEL")) {
kernel_specialization_level = (MetalPipelineType)atoi(envstr);
}
@@ -223,13 +181,6 @@ MetalDevice::MetalDevice(const DeviceInfo &info, Stats &stats, Profiler &profile
MetalDevice::~MetalDevice()
{
/* Cancel any async shader compilations that are in flight. */
cancel();
/* This lock safeguards against destruction during use (see other uses of
* existing_devices_mutex). */
thread_scoped_lock lock(existing_devices_mutex);
for (auto &tex : texture_slot_map) {
if (tex) {
[tex release];
@@ -327,21 +278,10 @@ void MetalDevice::make_source(MetalPipelineType pso_type, const uint kernel_feat
# define KERNEL_STRUCT_BEGIN(name, parent) \
string_replace_same_length(source, "kernel_data." #parent ".", "kernel_data_" #parent "_");
bool next_member_is_specialized = true;
# define KERNEL_STRUCT_MEMBER_DONT_SPECIALIZE next_member_is_specialized = false;
/* Add constants to md5 so that 'get_best_pipeline' is able to return a suitable match. */
# define KERNEL_STRUCT_MEMBER(parent, _type, name) \
if (next_member_is_specialized) { \
baked_constants += string(#parent "." #name "=") + \
to_string(_type(launch_params.data.parent.name)) + "\n"; \
} \
else { \
string_replace( \
source, "kernel_data_" #parent "_" #name, "kernel_data." #parent ".__unused_" #name); \
next_member_is_specialized = true; \
}
baked_constants += string(#parent "." #name "=") + \
to_string(_type(launch_params.data.parent.name)) + "\n";
# include "kernel/data_template.h"
@@ -386,66 +326,21 @@ bool MetalDevice::load_kernels(const uint _kernel_features)
* active, but may still need to be rendered without motion blur if that isn't active as well. */
motion_blur = kernel_features & KERNEL_FEATURE_OBJECT_MOTION;
/* Only request generic kernels if they aren't cached in memory. */
if (make_source_and_check_if_compile_needed(PSO_GENERIC)) {
/* If needed, load them asynchronously in order to responsively message progress to the user.
*/
int this_device_id = this->device_id;
auto compile_kernels_fn = ^() {
compile_and_load(this_device_id, PSO_GENERIC);
};
bool result = compile_and_load(PSO_GENERIC);
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0),
compile_kernels_fn);
}
return true;
reserve_local_memory(kernel_features);
return result;
}
bool MetalDevice::make_source_and_check_if_compile_needed(MetalPipelineType pso_type)
bool MetalDevice::compile_and_load(MetalPipelineType pso_type)
{
if (this->source[pso_type].empty()) {
make_source(pso_type, kernel_features);
make_source(pso_type, kernel_features);
if (!MetalDeviceKernels::should_load_kernels(this, pso_type)) {
/* We already have a full set of matching pipelines which are cached or queued. */
metal_printf("%s kernels already requested\n", kernel_type_as_string(pso_type));
return true;
}
return MetalDeviceKernels::should_load_kernels(this, pso_type);
}
void MetalDevice::compile_and_load(int device_id, MetalPipelineType pso_type)
{
/* Thread-safe front-end compilation. Typically the MSL->AIR compilation can take a few seconds,
* so we avoid blocking device tear-down if the user cancels a render immediately. */
id<MTLDevice> mtlDevice;
string source;
MetalGPUVendor device_vendor;
/* Safely gather any state required for the MSL->AIR compilation. */
{
thread_scoped_lock lock(existing_devices_mutex);
/* Check whether the device still exists. */
MetalDevice *instance = get_device_by_ID(device_id, lock);
if (!instance) {
metal_printf("Ignoring %s compilation request - device no longer exists\n",
kernel_type_as_string(pso_type));
return;
}
if (!instance->make_source_and_check_if_compile_needed(pso_type)) {
/* We already have a full set of matching pipelines which are cached or queued. Return early
* to avoid redundant MTLLibrary compilation. */
metal_printf("Ignoreing %s compilation request - kernels already requested\n",
kernel_type_as_string(pso_type));
return;
}
mtlDevice = instance->mtlDevice;
device_vendor = instance->device_vendor;
source = instance->source[pso_type];
}
/* Perform the actual compilation using our cached context. The MetalDevice can safely destruct
* in this time. */
MTLCompileOptions *options = [[MTLCompileOptions alloc] init];
@@ -464,15 +359,20 @@ void MetalDevice::compile_and_load(int device_id, MetalPipelineType pso_type)
if (getenv("CYCLES_METAL_PROFILING") || getenv("CYCLES_METAL_DEBUG")) {
path_write_text(path_cache_get(string_printf("%s.metal", kernel_type_as_string(pso_type))),
source);
source[pso_type]);
}
double starttime = time_dt();
const double starttime = time_dt();
NSError *error = NULL;
id<MTLLibrary> mtlLibrary = [mtlDevice newLibraryWithSource:@(source.c_str())
options:options
error:&error];
mtlLibrary[pso_type] = [mtlDevice newLibraryWithSource:@(source[pso_type].c_str())
options:options
error:&error];
if (!mtlLibrary[pso_type]) {
NSString *err = [error localizedDescription];
set_error(string_printf("Failed to compile library:\n%s", [err UTF8String]));
}
metal_printf("Front-end compilation finished in %.1f seconds (%s)\n",
time_dt() - starttime,
@@ -480,38 +380,17 @@ void MetalDevice::compile_and_load(int device_id, MetalPipelineType pso_type)
[options release];
bool blocking_pso_build = (getenv("CYCLES_METAL_PROFILING") ||
MetalDeviceKernels::is_benchmark_warmup());
if (blocking_pso_build) {
MetalDeviceKernels::wait_for_all();
starttime = 0.0;
}
return MetalDeviceKernels::load(this, pso_type);
}
/* Save the compiled MTLLibrary and trigger the AIR->PSO builds (if the MetalDevice still
* exists). */
{
thread_scoped_lock lock(existing_devices_mutex);
if (MetalDevice *instance = get_device_by_ID(device_id, lock)) {
if (mtlLibrary) {
instance->mtlLibrary[pso_type] = mtlLibrary;
void MetalDevice::reserve_local_memory(const uint kernel_features)
{
/* METAL_WIP - implement this */
}
starttime = time_dt();
MetalDeviceKernels::load(instance, pso_type);
}
else {
NSString *err = [error localizedDescription];
instance->set_error(string_printf("Failed to compile library:\n%s", [err UTF8String]));
}
}
}
if (starttime && blocking_pso_build) {
MetalDeviceKernels::wait_for_all();
metal_printf("Back-end compilation finished in %.1f seconds (%s)\n",
time_dt() - starttime,
kernel_type_as_string(pso_type));
}
void MetalDevice::init_host_memory()
{
/* METAL_WIP - implement this */
}
void MetalDevice::load_texture_info()
@@ -821,72 +700,55 @@ device_ptr MetalDevice::mem_alloc_sub_ptr(device_memory &mem, size_t offset, siz
return 0;
}
void MetalDevice::cancel()
{
/* Remove this device's ID from the list of active devices. Any pending compilation requests
* originating from this session will be cancelled. */
thread_scoped_lock lock(existing_devices_mutex);
if (device_id) {
active_device_ids.erase(device_id);
device_id = 0;
}
}
bool MetalDevice::is_ready(string &status) const
{
int num_loaded = MetalDeviceKernels::get_loaded_kernel_count(this, PSO_GENERIC);
if (num_loaded < DEVICE_KERNEL_NUM) {
status = string_printf("%d / %d render kernels loaded (may take a few minutes the first time)",
num_loaded,
DEVICE_KERNEL_NUM);
return false;
}
metal_printf("MetalDevice::is_ready(...) --> true\n");
return true;
}
void MetalDevice::optimize_for_scene(Scene *scene)
{
MetalPipelineType specialization_level = kernel_specialization_level;
if (!scene->params.background) {
/* In live viewport, don't specialize beyond intersection kernels for responsiveness. */
specialization_level = (MetalPipelineType)min(specialization_level, PSO_SPECIALIZED_INTERSECT);
if (specialization_level < PSO_SPECIALIZED_INTERSECT) {
return;
}
/* For responsive rendering, specialize the kernels in the background, and only if there isn't an
* existing "optimize_for_scene" request in flight. */
int this_device_id = this->device_id;
auto specialize_kernels_fn = ^() {
for (int level = 1; level <= int(specialization_level); level++) {
compile_and_load(this_device_id, MetalPipelineType(level));
}
/* PSO_SPECIALIZED_INTERSECT kernels are fast to specialize, so we always load them
* synchronously. */
compile_and_load(PSO_SPECIALIZED_INTERSECT);
if (specialization_level < PSO_SPECIALIZED_SHADE) {
return;
}
if (!scene->params.background) {
/* Don't load PSO_SPECIALIZED_SHADE kernels during viewport rendering as they are slower to
* build. */
return;
}
/* PSO_SPECIALIZED_SHADE kernels are slower to specialize, so we load them asynchronously, and
* only if there isn't an existing load in flight.
*/
auto specialize_shade_fn = ^() {
compile_and_load(PSO_SPECIALIZED_SHADE);
async_compile_and_load = false;
};
/* In normal use, we always compile the specialized kernels in the background. */
bool specialize_in_background = true;
bool async_specialize_shade = true;
/* Block if a per-kernel profiling is enabled (ensure steady rendering rate). */
if (getenv("CYCLES_METAL_PROFILING") != nullptr) {
specialize_in_background = false;
async_specialize_shade = false;
}
/* Block during benchmark warm-up to ensure kernels are cached prior to the observed run. */
if (MetalDeviceKernels::is_benchmark_warmup()) {
specialize_in_background = false;
}
if (specialize_in_background) {
if (!MetalDeviceKernels::any_specialization_happening_now()) {
if (async_specialize_shade) {
if (!async_compile_and_load) {
async_compile_and_load = true;
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0),
specialize_kernels_fn);
specialize_shade_fn);
}
else {
metal_printf("\"optimize_for_scene\" request already in flight - dropping request\n");
metal_printf(
"Async PSO_SPECIALIZED_SHADE load request already in progress - dropping request\n");
}
}
else {
specialize_kernels_fn();
specialize_shade_fn();
}
}

View File

@@ -64,8 +64,6 @@ struct MetalKernelPipeline {
void compile();
int originating_device_id;
id<MTLLibrary> mtlLibrary = nil;
MetalPipelineType pso_type;
string source_md5;
@@ -96,13 +94,9 @@ struct MetalKernelPipeline {
/* Cache of Metal kernels for each DeviceKernel. */
namespace MetalDeviceKernels {
bool any_specialization_happening_now();
int get_loaded_kernel_count(MetalDevice const *device, MetalPipelineType pso_type);
bool should_load_kernels(MetalDevice const *device, MetalPipelineType pso_type);
bool should_load_kernels(MetalDevice *device, MetalPipelineType pso_type);
bool load(MetalDevice *device, MetalPipelineType pso_type);
const MetalKernelPipeline *get_best_pipeline(const MetalDevice *device, DeviceKernel kernel);
void wait_for_all();
bool is_benchmark_warmup();
} /* namespace MetalDeviceKernels */

View File

@@ -49,18 +49,6 @@ struct ShaderCache {
if (MetalInfo::get_device_vendor(mtlDevice) == METAL_GPU_APPLE) {
switch (MetalInfo::get_apple_gpu_architecture(mtlDevice)) {
default:
case APPLE_M2_BIG:
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_COMPACT_SHADOW_STATES] = {384, 128};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_INIT_FROM_CAMERA] = {640, 128};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_INTERSECT_CLOSEST] = {1024, 64};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_INTERSECT_SHADOW] = {704, 704};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_INTERSECT_SUBSURFACE] = {640, 32};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_QUEUED_PATHS_ARRAY] = {896, 768};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_SHADE_BACKGROUND] = {512, 128};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_SHADE_SHADOW] = {32, 32};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_SHADE_SURFACE] = {768, 576};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_SORTED_PATHS_ARRAY] = {896, 768};
break;
case APPLE_M2:
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_COMPACT_SHADOW_STATES] = {32, 32};
occupancy_tuning[DEVICE_KERNEL_INTEGRATOR_INIT_FROM_CAMERA] = {832, 32};
@@ -98,17 +86,23 @@ struct ShaderCache {
void load_kernel(DeviceKernel kernel, MetalDevice *device, MetalPipelineType pso_type);
bool should_load_kernel(DeviceKernel device_kernel,
MetalDevice const *device,
MetalDevice *device,
MetalPipelineType pso_type);
void wait_for_all();
private:
friend ShaderCache *get_shader_cache(id<MTLDevice> mtlDevice);
void compile_thread_func(int thread_index);
using PipelineCollection = std::vector<unique_ptr<MetalKernelPipeline>>;
struct PipelineRequest {
MetalKernelPipeline *pipeline = nullptr;
std::function<void(MetalKernelPipeline *)> completionHandler;
};
struct OccupancyTuningParameters {
int threads_per_threadgroup = 0;
int num_threads_per_block = 0;
@@ -119,61 +113,35 @@ struct ShaderCache {
PipelineCollection pipelines[DEVICE_KERNEL_NUM];
id<MTLDevice> mtlDevice;
static bool running;
bool running = false;
std::condition_variable cond_var;
std::deque<MetalKernelPipeline *> request_queue;
std::deque<PipelineRequest> request_queue;
std::vector<std::thread> compile_threads;
std::atomic_int incomplete_requests = 0;
std::atomic_int incomplete_specialization_requests = 0;
};
bool ShaderCache::running = true;
const int MAX_POSSIBLE_GPUS_ON_SYSTEM = 8;
using DeviceShaderCache = std::pair<id<MTLDevice>, unique_ptr<ShaderCache>>;
int g_shaderCacheCount = 0;
DeviceShaderCache g_shaderCache[MAX_POSSIBLE_GPUS_ON_SYSTEM];
std::mutex g_shaderCacheMutex;
std::map<id<MTLDevice>, unique_ptr<ShaderCache>> g_shaderCache;
ShaderCache *get_shader_cache(id<MTLDevice> mtlDevice)
{
for (int i = 0; i < g_shaderCacheCount; i++) {
if (g_shaderCache[i].first == mtlDevice) {
return g_shaderCache[i].second.get();
}
thread_scoped_lock lock(g_shaderCacheMutex);
auto it = g_shaderCache.find(mtlDevice);
if (it != g_shaderCache.end()) {
return it->second.get();
}
static thread_mutex g_shaderCacheCountMutex;
g_shaderCacheCountMutex.lock();
int index = g_shaderCacheCount++;
g_shaderCacheCountMutex.unlock();
assert(index < MAX_POSSIBLE_GPUS_ON_SYSTEM);
g_shaderCache[index].first = mtlDevice;
g_shaderCache[index].second = make_unique<ShaderCache>(mtlDevice);
return g_shaderCache[index].second.get();
g_shaderCache[mtlDevice] = make_unique<ShaderCache>(mtlDevice);
return g_shaderCache[mtlDevice].get();
}
ShaderCache::~ShaderCache()
{
metal_printf("ShaderCache shutting down with incomplete_requests = %d\n",
int(incomplete_requests));
running = false;
cond_var.notify_all();
int num_incomplete = int(incomplete_requests);
if (num_incomplete) {
/* Shutting down the app with incomplete shader compilation requests. Give 1 second's grace for
* clean shutdown. */
metal_printf("ShaderCache busy (incomplete_requests = %d)...\n", num_incomplete);
std::this_thread::sleep_for(std::chrono::seconds(1));
num_incomplete = int(incomplete_requests);
}
if (num_incomplete && !MetalDeviceKernels::is_benchmark_warmup()) {
metal_printf("ShaderCache still busy (incomplete_requests = %d). Terminating...\n",
num_incomplete);
std::terminate();
}
metal_printf("ShaderCache idle. Shutting down.\n");
for (auto &thread : compile_threads) {
thread.join();
}
@@ -188,69 +156,35 @@ void ShaderCache::wait_for_all()
void ShaderCache::compile_thread_func(int thread_index)
{
while (running) {
while (1) {
/* wait for / acquire next request */
MetalKernelPipeline *pipeline;
PipelineRequest request;
{
thread_scoped_lock lock(cache_mutex);
cond_var.wait(lock, [&] { return !running || !request_queue.empty(); });
if (!running || request_queue.empty()) {
continue;
if (!running) {
break;
}
pipeline = request_queue.front();
request_queue.pop_front();
}
/* Service the request. */
DeviceKernel device_kernel = pipeline->device_kernel;
MetalPipelineType pso_type = pipeline->pso_type;
if (MetalDevice::is_device_cancelled(pipeline->originating_device_id)) {
/* The originating MetalDevice is no longer active, so this request is obsolete. */
metal_printf("Cancelling compilation of %s (%s)\n",
device_kernel_as_string(device_kernel),
kernel_type_as_string(pso_type));
}
else {
/* Do the actual compilation. */
pipeline->compile();
thread_scoped_lock lock(cache_mutex);
auto &collection = pipelines[device_kernel];
/* Cache up to 3 kernel variants with the same pso_type in memory, purging oldest first. */
int max_entries_of_same_pso_type = 3;
for (int i = (int)collection.size() - 1; i >= 0; i--) {
if (collection[i]->pso_type == pso_type) {
max_entries_of_same_pso_type -= 1;
if (max_entries_of_same_pso_type == 0) {
metal_printf("Purging oldest %s:%s kernel from ShaderCache\n",
kernel_type_as_string(pso_type),
device_kernel_as_string(device_kernel));
collection.erase(collection.begin() + i);
break;
}
}
if (!request_queue.empty()) {
request = request_queue.front();
request_queue.pop_front();
}
collection.push_back(unique_ptr<MetalKernelPipeline>(pipeline));
}
incomplete_requests--;
if (pso_type != PSO_GENERIC) {
incomplete_specialization_requests--;
/* service request */
if (request.pipeline) {
request.pipeline->compile();
incomplete_requests--;
}
}
}
bool ShaderCache::should_load_kernel(DeviceKernel device_kernel,
MetalDevice const *device,
MetalDevice *device,
MetalPipelineType pso_type)
{
if (!running) {
return false;
}
if (device_kernel == DEVICE_KERNEL_INTEGRATOR_MEGAKERNEL) {
/* Skip megakernel. */
return false;
@@ -306,6 +240,7 @@ void ShaderCache::load_kernel(DeviceKernel device_kernel,
/* create compiler threads on first run */
thread_scoped_lock lock(cache_mutex);
if (compile_threads.empty()) {
running = true;
for (int i = 0; i < max_mtlcompiler_threads; i++) {
compile_threads.push_back(std::thread([&] { compile_thread_func(i); }));
}
@@ -317,43 +252,65 @@ void ShaderCache::load_kernel(DeviceKernel device_kernel,
}
incomplete_requests++;
if (pso_type != PSO_GENERIC) {
incomplete_specialization_requests++;
}
MetalKernelPipeline *pipeline = new MetalKernelPipeline;
/* Keep track of the originating device's ID so that we can cancel requests if the device ceases
* to be active. */
pipeline->originating_device_id = device->device_id;
memcpy(&pipeline->kernel_data_, &device->launch_params.data, sizeof(pipeline->kernel_data_));
pipeline->pso_type = pso_type;
pipeline->mtlDevice = mtlDevice;
pipeline->source_md5 = device->source_md5[pso_type];
pipeline->mtlLibrary = device->mtlLibrary[pso_type];
pipeline->device_kernel = device_kernel;
pipeline->threads_per_threadgroup = device->max_threads_per_threadgroup;
PipelineRequest request;
request.pipeline = new MetalKernelPipeline;
memcpy(&request.pipeline->kernel_data_,
&device->launch_params.data,
sizeof(request.pipeline->kernel_data_));
request.pipeline->pso_type = pso_type;
request.pipeline->mtlDevice = mtlDevice;
request.pipeline->source_md5 = device->source_md5[pso_type];
request.pipeline->mtlLibrary = device->mtlLibrary[pso_type];
request.pipeline->device_kernel = device_kernel;
request.pipeline->threads_per_threadgroup = device->max_threads_per_threadgroup;
if (occupancy_tuning[device_kernel].threads_per_threadgroup) {
pipeline->threads_per_threadgroup = occupancy_tuning[device_kernel].threads_per_threadgroup;
pipeline->num_threads_per_block = occupancy_tuning[device_kernel].num_threads_per_block;
request.pipeline->threads_per_threadgroup =
occupancy_tuning[device_kernel].threads_per_threadgroup;
request.pipeline->num_threads_per_block =
occupancy_tuning[device_kernel].num_threads_per_block;
}
/* metalrt options */
pipeline->use_metalrt = device->use_metalrt;
pipeline->metalrt_features = device->use_metalrt ?
(device->kernel_features & METALRT_FEATURE_MASK) :
0;
request.pipeline->use_metalrt = device->use_metalrt;
request.pipeline->metalrt_features = device->use_metalrt ?
(device->kernel_features & METALRT_FEATURE_MASK) :
0;
{
thread_scoped_lock lock(cache_mutex);
request_queue.push_back(pipeline);
auto &collection = pipelines[device_kernel];
/* Cache up to 3 kernel variants with the same pso_type, purging oldest first. */
int max_entries_of_same_pso_type = 3;
for (int i = (int)collection.size() - 1; i >= 0; i--) {
if (collection[i]->pso_type == pso_type) {
max_entries_of_same_pso_type -= 1;
if (max_entries_of_same_pso_type == 0) {
metal_printf("Purging oldest %s:%s kernel from ShaderCache\n",
kernel_type_as_string(pso_type),
device_kernel_as_string(device_kernel));
collection.erase(collection.begin() + i);
break;
}
}
}
collection.push_back(unique_ptr<MetalKernelPipeline>(request.pipeline));
request_queue.push_back(request);
}
cond_var.notify_one();
}
MetalKernelPipeline *ShaderCache::get_best_pipeline(DeviceKernel kernel, const MetalDevice *device)
{
thread_scoped_lock lock(cache_mutex);
auto &collection = pipelines[kernel];
if (collection.empty()) {
return nullptr;
}
/* metalrt options */
bool use_metalrt = device->use_metalrt;
bool device_metalrt_hair = use_metalrt && device->kernel_features & KERNEL_FEATURE_HAIR;
@@ -365,43 +322,34 @@ MetalKernelPipeline *ShaderCache::get_best_pipeline(DeviceKernel kernel, const M
device->kernel_features & KERNEL_FEATURE_OBJECT_MOTION;
MetalKernelPipeline *best_pipeline = nullptr;
while (!best_pipeline) {
{
thread_scoped_lock lock(cache_mutex);
for (auto &pipeline : pipelines[kernel]) {
if (!pipeline->loaded) {
/* still loading - ignore */
continue;
}
bool pipeline_metalrt_hair = pipeline->metalrt_features & KERNEL_FEATURE_HAIR;
bool pipeline_metalrt_hair_thick = pipeline->metalrt_features & KERNEL_FEATURE_HAIR_THICK;
bool pipeline_metalrt_pointcloud = pipeline->metalrt_features & KERNEL_FEATURE_POINTCLOUD;
bool pipeline_metalrt_motion = use_metalrt &&
pipeline->metalrt_features & KERNEL_FEATURE_OBJECT_MOTION;
if (pipeline->use_metalrt != use_metalrt || pipeline_metalrt_hair != device_metalrt_hair ||
pipeline_metalrt_hair_thick != device_metalrt_hair_thick ||
pipeline_metalrt_pointcloud != device_metalrt_pointcloud ||
pipeline_metalrt_motion != device_metalrt_motion) {
/* wrong combination of metalrt options */
continue;
}
if (pipeline->pso_type != PSO_GENERIC) {
if (pipeline->source_md5 == device->source_md5[PSO_SPECIALIZED_INTERSECT] ||
pipeline->source_md5 == device->source_md5[PSO_SPECIALIZED_SHADE]) {
best_pipeline = pipeline.get();
}
}
else if (!best_pipeline) {
best_pipeline = pipeline.get();
}
}
for (auto &pipeline : collection) {
if (!pipeline->loaded) {
/* still loading - ignore */
continue;
}
if (!best_pipeline) {
std::this_thread::sleep_for(std::chrono::milliseconds(100));
bool pipeline_metalrt_hair = pipeline->metalrt_features & KERNEL_FEATURE_HAIR;
bool pipeline_metalrt_hair_thick = pipeline->metalrt_features & KERNEL_FEATURE_HAIR_THICK;
bool pipeline_metalrt_pointcloud = pipeline->metalrt_features & KERNEL_FEATURE_POINTCLOUD;
bool pipeline_metalrt_motion = use_metalrt &&
pipeline->metalrt_features & KERNEL_FEATURE_OBJECT_MOTION;
if (pipeline->use_metalrt != use_metalrt || pipeline_metalrt_hair != device_metalrt_hair ||
pipeline_metalrt_hair_thick != device_metalrt_hair_thick ||
pipeline_metalrt_pointcloud != device_metalrt_pointcloud ||
pipeline_metalrt_motion != device_metalrt_motion) {
/* wrong combination of metalrt options */
continue;
}
if (pipeline->pso_type != PSO_GENERIC) {
if (pipeline->source_md5 == device->source_md5[PSO_SPECIALIZED_INTERSECT] ||
pipeline->source_md5 == device->source_md5[PSO_SPECIALIZED_SHADE]) {
best_pipeline = pipeline.get();
}
}
else if (!best_pipeline) {
best_pipeline = pipeline.get();
}
}
@@ -460,18 +408,13 @@ static MTLFunctionConstantValues *GetConstantValues(KernelData const *data = nul
if (!data) {
data = &zero_data;
}
[constant_values setConstantValue:&zero_data type:MTLDataType_int atIndex:Kernel_DummyConstant];
bool next_member_is_specialized = true;
# define KERNEL_STRUCT_MEMBER_DONT_SPECIALIZE next_member_is_specialized = false;
int zero_int = 0;
[constant_values setConstantValue:&zero_int type:MTLDataType_int atIndex:Kernel_DummyConstant];
# define KERNEL_STRUCT_MEMBER(parent, _type, name) \
[constant_values setConstantValue:next_member_is_specialized ? (void *)&data->parent.name : \
(void *)&zero_data \
[constant_values setConstantValue:&data->parent.name \
type:MTLDataType_##_type \
atIndex:KernelData_##parent##_##name]; \
next_member_is_specialized = true;
atIndex:KernelData_##parent##_##name];
# include "kernel/data_template.h"
@@ -721,60 +664,51 @@ void MetalKernelPipeline::compile()
double starttime = time_dt();
/* Block on load to ensure we continue with a valid kernel function */
if (creating_new_archive) {
starttime = time_dt();
NSError *error;
if (![archive addComputePipelineFunctionsWithDescriptor:computePipelineStateDescriptor
error:&error]) {
MTLNewComputePipelineStateWithReflectionCompletionHandler completionHandler = ^(
id<MTLComputePipelineState> computePipelineState,
MTLComputePipelineReflection *reflection,
NSError *error) {
bool recreate_archive = false;
if (computePipelineState == nil && archive) {
NSString *errStr = [error localizedDescription];
metal_printf("Failed to add PSO to archive:\n%s\n", errStr ? [errStr UTF8String] : "nil");
metal_printf(
"Failed to create compute pipeline state \"%s\" from archive - attempting recreation... "
"(error: %s)\n",
device_kernel_as_string((DeviceKernel)device_kernel),
errStr ? [errStr UTF8String] : "nil");
computePipelineState = [mtlDevice
newComputePipelineStateWithDescriptor:computePipelineStateDescriptor
options:MTLPipelineOptionNone
reflection:nullptr
error:&error];
recreate_archive = true;
}
}
pipeline = [mtlDevice newComputePipelineStateWithDescriptor:computePipelineStateDescriptor
options:pipelineOptions
reflection:nullptr
error:&error];
double duration = time_dt() - starttime;
bool recreate_archive = false;
if (pipeline == nil && archive) {
NSString *errStr = [error localizedDescription];
metal_printf(
"Failed to create compute pipeline state \"%s\" from archive - attempting recreation... "
"(error: %s)\n",
device_kernel_as_string((DeviceKernel)device_kernel),
errStr ? [errStr UTF8String] : "nil");
pipeline = [mtlDevice newComputePipelineStateWithDescriptor:computePipelineStateDescriptor
options:MTLPipelineOptionNone
reflection:nullptr
error:&error];
recreate_archive = true;
}
if (computePipelineState == nil) {
NSString *errStr = [error localizedDescription];
error_str = string_printf("Failed to create compute pipeline state \"%s\", error: \n",
device_kernel_as_string((DeviceKernel)device_kernel));
error_str += (errStr ? [errStr UTF8String] : "nil");
metal_printf("%16s | %2d | %-55s | %7.2fs | FAILED!\n",
kernel_type_as_string(pso_type),
device_kernel,
device_kernel_as_string((DeviceKernel)device_kernel),
duration);
return;
}
double duration = time_dt() - starttime;
if (!num_threads_per_block) {
num_threads_per_block = round_down(computePipelineState.maxTotalThreadsPerThreadgroup,
computePipelineState.threadExecutionWidth);
num_threads_per_block = std::max(num_threads_per_block,
(int)computePipelineState.threadExecutionWidth);
}
if (pipeline == nil) {
NSString *errStr = [error localizedDescription];
error_str = string_printf("Failed to create compute pipeline state \"%s\", error: \n",
device_kernel_as_string((DeviceKernel)device_kernel));
error_str += (errStr ? [errStr UTF8String] : "nil");
metal_printf("%16s | %2d | %-55s | %7.2fs | FAILED!\n",
kernel_type_as_string(pso_type),
device_kernel,
device_kernel_as_string((DeviceKernel)device_kernel),
duration);
return;
}
this->pipeline = computePipelineState;
if (!num_threads_per_block) {
num_threads_per_block = round_down(pipeline.maxTotalThreadsPerThreadgroup,
pipeline.threadExecutionWidth);
num_threads_per_block = std::max(num_threads_per_block, (int)pipeline.threadExecutionWidth);
}
if (@available(macOS 11.0, *)) {
if (ShaderCache::running) {
if (@available(macOS 11.0, *)) {
if (creating_new_archive || recreate_archive) {
if (![archive serializeToURL:[NSURL fileURLWithPath:@(metalbin_path.c_str())]
error:&error]) {
@@ -786,7 +720,24 @@ void MetalKernelPipeline::compile()
}
}
}
};
/* Block on load to ensure we continue with a valid kernel function */
if (creating_new_archive) {
starttime = time_dt();
NSError *error;
if (![archive addComputePipelineFunctionsWithDescriptor:computePipelineStateDescriptor
error:&error]) {
NSString *errStr = [error localizedDescription];
metal_printf("Failed to add PSO to archive:\n%s\n", errStr ? [errStr UTF8String] : "nil");
}
}
id<MTLComputePipelineState> pipeline = [mtlDevice
newComputePipelineStateWithDescriptor:computePipelineStateDescriptor
options:pipelineOptions
reflection:nullptr
error:&error];
completionHandler(pipeline, nullptr, error);
this->loaded = true;
[computePipelineStateDescriptor release];
@@ -812,6 +763,8 @@ void MetalKernelPipeline::compile()
}
}
double duration = time_dt() - starttime;
if (!use_binary_archive) {
metal_printf("%16s | %2d | %-55s | %7.2fs\n",
kernel_type_as_string(pso_type),
@@ -832,69 +785,36 @@ void MetalKernelPipeline::compile()
bool MetalDeviceKernels::load(MetalDevice *device, MetalPipelineType pso_type)
{
const double starttime = time_dt();
auto shader_cache = get_shader_cache(device->mtlDevice);
for (int i = 0; i < DEVICE_KERNEL_NUM; i++) {
shader_cache->load_kernel((DeviceKernel)i, device, pso_type);
}
shader_cache->wait_for_all();
metal_printf("Back-end compilation finished in %.1f seconds (%s)\n",
time_dt() - starttime,
kernel_type_as_string(pso_type));
return true;
}
void MetalDeviceKernels::wait_for_all()
bool MetalDeviceKernels::should_load_kernels(MetalDevice *device, MetalPipelineType pso_type)
{
for (int i = 0; i < g_shaderCacheCount; i++) {
g_shaderCache[i].second->wait_for_all();
}
}
bool MetalDeviceKernels::any_specialization_happening_now()
{
/* Return true if any ShaderCaches have ongoing specialization requests (typically there will be
* only 1). */
for (int i = 0; i < g_shaderCacheCount; i++) {
if (g_shaderCache[i].second->incomplete_specialization_requests > 0) {
auto shader_cache = get_shader_cache(device->mtlDevice);
for (int i = 0; i < DEVICE_KERNEL_NUM; i++) {
if (shader_cache->should_load_kernel((DeviceKernel)i, device, pso_type)) {
return true;
}
}
return false;
}
int MetalDeviceKernels::get_loaded_kernel_count(MetalDevice const *device,
MetalPipelineType pso_type)
{
auto shader_cache = get_shader_cache(device->mtlDevice);
int loaded_count = DEVICE_KERNEL_NUM;
for (int i = 0; i < DEVICE_KERNEL_NUM; i++) {
if (shader_cache->should_load_kernel((DeviceKernel)i, device, pso_type)) {
loaded_count -= 1;
}
}
return loaded_count;
}
bool MetalDeviceKernels::should_load_kernels(MetalDevice const *device, MetalPipelineType pso_type)
{
return get_loaded_kernel_count(device, pso_type) != DEVICE_KERNEL_NUM;
}
const MetalKernelPipeline *MetalDeviceKernels::get_best_pipeline(const MetalDevice *device,
DeviceKernel kernel)
{
return get_shader_cache(device->mtlDevice)->get_best_pipeline(kernel, device);
}
bool MetalDeviceKernels::is_benchmark_warmup()
{
NSArray *args = [[NSProcessInfo processInfo] arguments];
for (int i = 0; i < args.count; i++) {
if (const char *arg = [[args objectAtIndex:i] cStringUsingEncoding:NSASCIIStringEncoding]) {
if (!strcmp(arg, "--warm-up")) {
return true;
}
}
}
return false;
}
CCL_NAMESPACE_END
#endif /* WITH_METAL*/

View File

@@ -202,9 +202,6 @@ MetalDeviceQueue::~MetalDeviceQueue()
assert(mtlCommandBuffer_ == nil);
assert(command_buffers_submitted_ == command_buffers_completed_);
close_compute_encoder();
close_blit_encoder();
if (@available(macos 10.14, *)) {
[shared_event_listener_ release];
[shared_event_ release];
@@ -278,8 +275,7 @@ int MetalDeviceQueue::num_concurrent_states(const size_t state_size) const
if (metal_device_->device_vendor == METAL_GPU_APPLE) {
result *= 4;
/* Increasing the state count doesn't notably benefit M1-family systems. */
if (MetalInfo::get_apple_gpu_architecture(metal_device_->mtlDevice) != APPLE_M1) {
if (MetalInfo::get_apple_gpu_architecture(metal_device_->mtlDevice) == APPLE_M2) {
size_t system_ram = system_physical_ram();
size_t allocated_so_far = [metal_device_->mtlDevice currentAllocatedSize];
size_t max_recommended_working_set = [metal_device_->mtlDevice recommendedMaxWorkingSetSize];
@@ -641,7 +637,9 @@ bool MetalDeviceQueue::synchronize()
return false;
}
close_compute_encoder();
if (mtlComputeEncoder_) {
close_compute_encoder();
}
close_blit_encoder();
if (mtlCommandBuffer_) {
@@ -704,10 +702,6 @@ bool MetalDeviceQueue::synchronize()
void MetalDeviceQueue::zero_to_device(device_memory &mem)
{
if (metal_device_->have_error()) {
return;
}
assert(mem.type != MEM_GLOBAL && mem.type != MEM_TEXTURE);
if (mem.memory_size() == 0) {
@@ -735,10 +729,6 @@ void MetalDeviceQueue::zero_to_device(device_memory &mem)
void MetalDeviceQueue::copy_to_device(device_memory &mem)
{
if (metal_device_->have_error()) {
return;
}
if (mem.memory_size() == 0) {
return;
}
@@ -781,10 +771,6 @@ void MetalDeviceQueue::copy_to_device(device_memory &mem)
void MetalDeviceQueue::copy_from_device(device_memory &mem)
{
if (metal_device_->have_error()) {
return;
}
assert(mem.type != MEM_GLOBAL && mem.type != MEM_TEXTURE);
if (mem.memory_size() == 0) {
@@ -857,7 +843,9 @@ id<MTLComputeCommandEncoder> MetalDeviceQueue::get_compute_encoder(DeviceKernel
if (@available(macos 10.14, *)) {
if (timing_shared_event_) {
/* Close the current encoder to ensure we're able to capture per-encoder timing data. */
close_compute_encoder();
if (mtlComputeEncoder_) {
close_compute_encoder();
}
}
if (mtlComputeEncoder_) {
@@ -897,7 +885,9 @@ id<MTLBlitCommandEncoder> MetalDeviceQueue::get_blit_encoder()
return mtlBlitEncoder_;
}
close_compute_encoder();
if (mtlComputeEncoder_) {
close_compute_encoder();
}
if (!mtlCommandBuffer_) {
mtlCommandBuffer_ = [mtlCommandQueue_ commandBuffer];
@@ -911,14 +901,12 @@ id<MTLBlitCommandEncoder> MetalDeviceQueue::get_blit_encoder()
void MetalDeviceQueue::close_compute_encoder()
{
if (mtlComputeEncoder_) {
[mtlComputeEncoder_ endEncoding];
mtlComputeEncoder_ = nil;
[mtlComputeEncoder_ endEncoding];
mtlComputeEncoder_ = nil;
if (@available(macos 10.14, *)) {
if (timing_shared_event_) {
[mtlCommandBuffer_ encodeSignalEvent:timing_shared_event_ value:timing_shared_event_id_++];
}
if (@available(macos 10.14, *)) {
if (timing_shared_event_) {
[mtlCommandBuffer_ encodeSignalEvent:timing_shared_event_ value:timing_shared_event_id_++];
}
}
}

View File

@@ -29,7 +29,6 @@ enum AppleGPUArchitecture {
APPLE_UNKNOWN,
APPLE_M1,
APPLE_M2,
APPLE_M2_BIG,
};
/* Contains static Metal helper functions. */

View File

@@ -52,7 +52,7 @@ AppleGPUArchitecture MetalInfo::get_apple_gpu_architecture(id<MTLDevice> device)
return APPLE_M1;
}
else if (strstr(device_name, "M2")) {
return get_apple_gpu_core_count(device) <= 10 ? APPLE_M2 : APPLE_M2_BIG;
return APPLE_M2;
}
return APPLE_UNKNOWN;
}

View File

@@ -429,12 +429,7 @@ void OneapiDevice::check_usm(SyclQueue *queue_, const void *usm_ptr, bool allow_
queue->get_device().get_info<sycl::info::device::device_type>();
sycl::usm::alloc usm_type = get_pointer_type(usm_ptr, queue->get_context());
(void)usm_type;
# ifndef WITH_ONEAPI_SYCL_HOST_TASK
const sycl::usm::alloc main_memory_type = sycl::usm::alloc::device;
# else
const sycl::usm::alloc main_memory_type = sycl::usm::alloc::host;
# endif
assert(usm_type == main_memory_type ||
assert(usm_type == sycl::usm::alloc::device ||
(usm_type == sycl::usm::alloc::host &&
(allow_host || device_type == sycl::info::device_type::cpu)) ||
usm_type == sycl::usm::alloc::unknown);
@@ -483,11 +478,7 @@ void *OneapiDevice::usm_alloc_device(SyclQueue *queue_, size_t memory_size)
{
assert(queue_);
sycl::queue *queue = reinterpret_cast<sycl::queue *>(queue_);
# ifndef WITH_ONEAPI_SYCL_HOST_TASK
return sycl::malloc_device(memory_size, *queue);
# else
return sycl::malloc_host(memory_size, *queue);
# endif
}
void OneapiDevice::usm_free(SyclQueue *queue_, void *usm_ptr)
@@ -745,11 +736,7 @@ char *OneapiDevice::device_capabilities()
const std::vector<sycl::device> &oneapi_devices = available_devices();
for (const sycl::device &device : oneapi_devices) {
# ifndef WITH_ONEAPI_SYCL_HOST_TASK
const std::string &name = device.get_info<sycl::info::device::name>();
# else
const std::string &name = "SYCL Host Task (Debug)";
# endif
capabilities << std::string("\t") << name << "\n";
# define WRITE_ATTR(attribute_name, attribute_variable) \
@@ -826,11 +813,7 @@ void OneapiDevice::iterate_devices(OneAPIDeviceIteratorCallback cb, void *user_p
for (sycl::device &device : devices) {
const std::string &platform_name =
device.get_platform().get_info<sycl::info::platform::name>();
# ifndef WITH_ONEAPI_SYCL_HOST_TASK
std::string name = device.get_info<sycl::info::device::name>();
# else
std::string name = "SYCL Host Task (Debug)";
# endif
std::string id = "ONEAPI_" + platform_name + "_" + name;
if (device.has(sycl::aspect::ext_intel_pci_address)) {
id.append("_" + device.get_info<sycl::ext::intel::info::device::pci_address>());

View File

@@ -579,11 +579,7 @@ bool OptiXDevice::load_kernels(const uint kernel_features)
link_options.maxTraceDepth = 1;
link_options.debugLevel = module_options.debugLevel;
if (use_osl) {
/* Re-create OSL pipeline in case kernels are reloaded after it has been created before. */
load_osl_kernels();
}
else if (kernel_features & (KERNEL_FEATURE_NODE_RAYTRACE | KERNEL_FEATURE_MNEE)) {
if (kernel_features & (KERNEL_FEATURE_NODE_RAYTRACE | KERNEL_FEATURE_MNEE) && !use_osl) {
/* Create shader raytracing and MNEE pipeline. */
vector<OptixProgramGroup> pipeline_groups;
pipeline_groups.reserve(NUM_PROGRAM_GROUPS);
@@ -747,11 +743,6 @@ bool OptiXDevice::load_osl_kernels()
}
}
if (osl_kernels.empty()) {
/* No OSL shader groups, so no need to create a pipeline. */
return true;
}
OptixProgramGroupOptions group_options = {}; /* There are no options currently. */
OptixModuleCompileOptions module_options = {};
module_options.optLevel = OPTIX_COMPILE_OPTIMIZATION_LEVEL_3;
@@ -866,11 +857,6 @@ bool OptiXDevice::load_osl_kernels()
optix_assert(optixSbtRecordPackHeader(osl_groups[i], &sbt_data[NUM_PROGRAM_GROUPS + i]));
optix_assert(optixProgramGroupGetStackSize(osl_groups[i], &osl_stack_size[i]));
}
else {
/* Default to "__direct_callable__dummy_services", so that OSL evaluation for empty
* materials has direct callables to call and does not crash. */
optix_assert(optixSbtRecordPackHeader(osl_groups.back(), &sbt_data[NUM_PROGRAM_GROUPS + i]));
}
}
sbt_data.copy_to_device(); /* Upload updated SBT to device. */

View File

@@ -10,4 +10,4 @@ set(LICENSES
Zlib-license.txt
)
delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${LICENSES}" ${CYCLES_INSTALL_PATH}/license)
delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${LICENSES}" ${CYCLES_INSTALL_PATH}/license)

View File

@@ -5,9 +5,6 @@ set(INC
..
)
set(INC_SYS
)
set(SRC
node.cpp
node_type.cpp

View File

@@ -26,7 +26,8 @@ class HdCyclesVolumeLoader : public VDBImageLoader {
HdCyclesVolumeLoader(const std::string &filePath, const std::string &gridName)
: VDBImageLoader(gridName)
{
/* Disable delay loading and file copying, this has poor performance on network drivers. */
/* Disably delay loading and file copying, this has poor performance
* on network drivers. */
const bool delay_load = false;
openvdb::io::File file(filePath);
file.setCopyMaxBytes(0);

View File

@@ -5,9 +5,6 @@ set(INC
..
)
set(INC_SYS
)
set(SRC
adaptive_sampling.cpp
denoiser.cpp

View File

@@ -390,9 +390,6 @@ void PathTrace::path_trace(RenderWork &render_work)
const int num_samples = render_work.path_trace.num_samples;
PathTraceWork *path_trace_work = path_trace_works_[i].get();
if (path_trace_work->get_device()->have_error()) {
return;
}
PathTraceWork::RenderStatistics statistics;
path_trace_work->render_samples(statistics,

View File

@@ -14,7 +14,9 @@ set(INC_SYS
set(SRC_KERNEL_DEVICE_CPU
device/cpu/kernel.cpp
device/cpu/kernel_sse2.cpp
device/cpu/kernel_sse3.cpp
device/cpu/kernel_sse41.cpp
device/cpu/kernel_avx.cpp
device/cpu/kernel_avx2.cpp
)
@@ -726,55 +728,50 @@ if(WITH_CYCLES_DEVICE_ONEAPI)
${SRC_UTIL_HEADERS}
)
set(SYCL_OFFLINE_COMPILER_PARALLEL_JOBS 1 CACHE STRING "Number of parallel compiler instances to use for device binaries compilation (expect ~8GB peak memory usage per instance).")
if(WITH_CYCLES_ONEAPI_BINARIES)
set (SYCL_OFFLINE_COMPILER_PARALLEL_JOBS 1 CACHE STRING "Number of parallel compiler instances to use for device binaries compilation (expect ~8GB peak memory usage per instance).")
if (WITH_CYCLES_ONEAPI_BINARIES)
message(STATUS "${SYCL_OFFLINE_COMPILER_PARALLEL_JOBS} instance(s) of oneAPI offline compiler will be used.")
endif()
# SYCL_CPP_FLAGS is a variable that the user can set to pass extra compiler options
set(sycl_compiler_flags
${CMAKE_CURRENT_SOURCE_DIR}/${SRC_KERNEL_DEVICE_ONEAPI}
-fsycl
-fsycl-unnamed-lambda
-fdelayed-template-parsing
-mllvm -inlinedefault-threshold=250
-mllvm -inlinehint-threshold=350
-fsycl-device-code-split=per_kernel
-fsycl-max-parallel-link-jobs=${SYCL_OFFLINE_COMPILER_PARALLEL_JOBS}
-shared
-DWITH_ONEAPI
-ffast-math
-DNDEBUG
-O2
-o ${cycles_kernel_oneapi_lib}
-I${CMAKE_CURRENT_SOURCE_DIR}/..
${SYCL_CPP_FLAGS}
)
if(WITH_CYCLES_ONEAPI_HOST_TASK_EXECUTION)
list(APPEND sycl_compiler_flags -DWITH_ONEAPI_SYCL_HOST_TASK)
endif()
${CMAKE_CURRENT_SOURCE_DIR}/${SRC_KERNEL_DEVICE_ONEAPI}
-fsycl
-fsycl-unnamed-lambda
-fdelayed-template-parsing
-mllvm -inlinedefault-threshold=250
-mllvm -inlinehint-threshold=350
-fsycl-device-code-split=per_kernel
-fsycl-max-parallel-link-jobs=${SYCL_OFFLINE_COMPILER_PARALLEL_JOBS}
-shared
-DWITH_ONEAPI
-ffast-math
-DNDEBUG
-O2
-o ${cycles_kernel_oneapi_lib}
-I${CMAKE_CURRENT_SOURCE_DIR}/..
${SYCL_CPP_FLAGS}
)
# Set defaults for spir64 and spir64_gen options
if(NOT DEFINED CYCLES_ONEAPI_SYCL_OPTIONS_spir64)
if (NOT DEFINED CYCLES_ONEAPI_SYCL_OPTIONS_spir64)
set(CYCLES_ONEAPI_SYCL_OPTIONS_spir64 "-options '-ze-opt-large-register-file -ze-opt-regular-grf-kernel integrator_intersect'")
endif()
if(NOT DEFINED CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen)
set(CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen "${CYCLES_ONEAPI_SYCL_OPTIONS_spir64}" CACHE STRING "Extra build options for spir64_gen target")
if (NOT DEFINED CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen)
SET (CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen "${CYCLES_ONEAPI_SYCL_OPTIONS_spir64}" CACHE STRING "Extra build options for spir64_gen target")
endif()
# Enable zebin, a graphics binary format with improved compatibility.
string(PREPEND CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen "--format zebin ")
string(PREPEND CYCLES_ONEAPI_SYCL_OPTIONS_spir64_gen "-device ${CYCLES_ONEAPI_SPIR64_GEN_DEVICES} ")
# Host execution won't use GPU binaries, no need to compile them.
if(WITH_CYCLES_ONEAPI_BINARIES AND NOT WITH_CYCLES_ONEAPI_HOST_TASK_EXECUTION)
if (WITH_CYCLES_ONEAPI_BINARIES)
# AoT binaries aren't currently reused when calling sycl::build.
list(APPEND sycl_compiler_flags -DSYCL_SKIP_KERNELS_PRELOAD)
list (APPEND sycl_compiler_flags -DSYCL_SKIP_KERNELS_PRELOAD)
# Iterate over all targest and their options
list(JOIN CYCLES_ONEAPI_SYCL_TARGETS "," targets_string)
list(APPEND sycl_compiler_flags -fsycl-targets=${targets_string})
list (JOIN CYCLES_ONEAPI_SYCL_TARGETS "," targets_string)
list (APPEND sycl_compiler_flags -fsycl-targets=${targets_string})
foreach(target ${CYCLES_ONEAPI_SYCL_TARGETS})
if(DEFINED CYCLES_ONEAPI_SYCL_OPTIONS_${target})
list(APPEND sycl_compiler_flags -Xsycl-target-backend=${target} "${CYCLES_ONEAPI_SYCL_OPTIONS_${target}}")
list (APPEND sycl_compiler_flags -Xsycl-target-backend=${target} "${CYCLES_ONEAPI_SYCL_OPTIONS_${target}}")
endif()
endforeach()
else()
@@ -863,9 +860,8 @@ if(WITH_CYCLES_DEVICE_ONEAPI)
else()
list(APPEND sycl_compiler_flags -fPIC)
# We avoid getting __FAST_MATH__ to be defined when building on CentOS-7 until the compilation
# We avoid getting __FAST_MATH__ to be defined when building on CentOS 7 until the compilation
# crash it triggers at either AoT or JIT stages gets fixed.
# TODO: check if this is still needed on Rocky-8.
list(APPEND sycl_compiler_flags -fhonor-nans)
# add $ORIGIN to cycles_kernel_oneapi.so rpath so libsycl.so and
@@ -938,9 +934,14 @@ set_source_files_properties(device/cpu/kernel.cpp PROPERTIES COMPILE_FLAGS "${CY
if(CXX_HAS_SSE)
set_source_files_properties(device/cpu/kernel_sse2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS}")
set_source_files_properties(device/cpu/kernel_sse3.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE3_KERNEL_FLAGS}")
set_source_files_properties(device/cpu/kernel_sse41.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE41_KERNEL_FLAGS}")
endif()
if(CXX_HAS_AVX)
set_source_files_properties(device/cpu/kernel_avx.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX_KERNEL_FLAGS}")
endif()
if(CXX_HAS_AVX2)
set_source_files_properties(device/cpu/kernel_avx2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX2_KERNEL_FLAGS}")
endif()

View File

@@ -63,9 +63,8 @@ ccl_device void kernel_background_evaluate(KernelGlobals kg,
shader_setup_from_background(kg, &sd, ray_P, ray_D, ray_time);
/* Evaluate shader.
* This is being evaluated for all BSDFs, so path flag does not contain a specific type.
* However, we want to flag the ray visibility to ignore the sun in the background map. */
const uint32_t path_flag = PATH_RAY_EMISSION | PATH_RAY_IMPORTANCE_BAKE;
* This is being evaluated for all BSDFs, so path flag does not contain a specific type. */
const uint32_t path_flag = PATH_RAY_EMISSION;
surface_shader_eval<KERNEL_FEATURE_NODE_MASK_SURFACE_LIGHT &
~(KERNEL_FEATURE_NODE_RAYTRACE | KERNEL_FEATURE_NODE_LIGHT_PATH)>(
kg, INTEGRATOR_STATE_NULL, &sd, NULL, path_flag);

View File

@@ -102,9 +102,10 @@ ccl_device_inline float shift_cos_in(float cos_in, const float frequency_multipl
return val;
}
ccl_device_inline bool bsdf_is_transmission(ccl_private const ShaderClosure *sc, const float3 wo)
ccl_device_inline bool bsdf_is_transmission(ccl_private const ShaderClosure *sc,
const float3 omega_in)
{
return dot(sc->N, wo) < 0.0f;
return dot(sc->N, omega_in) < 0.0f;
}
ccl_device_inline int bsdf_sample(KernelGlobals kg,
@@ -113,7 +114,7 @@ ccl_device_inline int bsdf_sample(KernelGlobals kg,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness,
ccl_private float *eta)
@@ -125,43 +126,43 @@ ccl_device_inline int bsdf_sample(KernelGlobals kg,
switch (sc->type) {
case CLOSURE_BSDF_DIFFUSE_ID:
label = bsdf_diffuse_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_diffuse_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
#if defined(__SVM__) || defined(__OSL__)
case CLOSURE_BSDF_OREN_NAYAR_ID:
label = bsdf_oren_nayar_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_oren_nayar_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
# ifdef __OSL__
case CLOSURE_BSDF_PHONG_RAMP_ID:
label = bsdf_phong_ramp_sample(
sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness);
sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness);
*eta = 1.0f;
break;
case CLOSURE_BSDF_DIFFUSE_RAMP_ID:
label = bsdf_diffuse_ramp_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_diffuse_ramp_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
# endif
case CLOSURE_BSDF_TRANSLUCENT_ID:
label = bsdf_translucent_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_translucent_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
case CLOSURE_BSDF_REFLECTION_ID:
label = bsdf_reflection_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf, eta);
label = bsdf_reflection_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, eta);
*sampled_roughness = zero_float2();
break;
case CLOSURE_BSDF_REFRACTION_ID:
label = bsdf_refraction_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf, eta);
label = bsdf_refraction_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, eta);
*sampled_roughness = zero_float2();
break;
case CLOSURE_BSDF_TRANSPARENT_ID:
label = bsdf_transparent_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_transparent_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = zero_float2();
*eta = 1.0f;
break;
@@ -170,65 +171,85 @@ ccl_device_inline int bsdf_sample(KernelGlobals kg,
case CLOSURE_BSDF_MICROFACET_GGX_CLEARCOAT_ID:
case CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID:
label = bsdf_microfacet_ggx_sample(
kg, sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness, eta);
kg, sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness, eta);
break;
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID:
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_FRESNEL_ID:
label = bsdf_microfacet_multi_ggx_sample(
kg, sc, Ng, sd->wi, randu, randv, eval, wo, pdf, &sd->lcg_state, sampled_roughness, eta);
label = bsdf_microfacet_multi_ggx_sample(kg,
sc,
Ng,
sd->I,
randu,
randv,
eval,
omega_in,
pdf,
&sd->lcg_state,
sampled_roughness,
eta);
break;
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID:
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID:
label = bsdf_microfacet_multi_ggx_glass_sample(
kg, sc, Ng, sd->wi, randu, randv, eval, wo, pdf, &sd->lcg_state, sampled_roughness, eta);
label = bsdf_microfacet_multi_ggx_glass_sample(kg,
sc,
Ng,
sd->I,
randu,
randv,
eval,
omega_in,
pdf,
&sd->lcg_state,
sampled_roughness,
eta);
break;
case CLOSURE_BSDF_MICROFACET_BECKMANN_ID:
case CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID:
label = bsdf_microfacet_beckmann_sample(
kg, sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness, eta);
kg, sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness, eta);
break;
case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID:
label = bsdf_ashikhmin_shirley_sample(
sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness);
sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness);
*eta = 1.0f;
break;
case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID:
label = bsdf_ashikhmin_velvet_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_ashikhmin_velvet_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
case CLOSURE_BSDF_DIFFUSE_TOON_ID:
label = bsdf_diffuse_toon_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_diffuse_toon_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
case CLOSURE_BSDF_GLOSSY_TOON_ID:
label = bsdf_glossy_toon_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_glossy_toon_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
// double check if this is valid
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
case CLOSURE_BSDF_HAIR_REFLECTION_ID:
label = bsdf_hair_reflection_sample(
sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness);
sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness);
*eta = 1.0f;
break;
case CLOSURE_BSDF_HAIR_TRANSMISSION_ID:
label = bsdf_hair_transmission_sample(
sc, Ng, sd->wi, randu, randv, eval, wo, pdf, sampled_roughness);
sc, Ng, sd->I, randu, randv, eval, omega_in, pdf, sampled_roughness);
*eta = 1.0f;
break;
case CLOSURE_BSDF_HAIR_PRINCIPLED_ID:
label = bsdf_principled_hair_sample(
kg, sc, sd, randu, randv, eval, wo, pdf, sampled_roughness, eta);
kg, sc, sd, randu, randv, eval, omega_in, pdf, sampled_roughness, eta);
break;
case CLOSURE_BSDF_PRINCIPLED_DIFFUSE_ID:
label = bsdf_principled_diffuse_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_principled_diffuse_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
case CLOSURE_BSDF_PRINCIPLED_SHEEN_ID:
label = bsdf_principled_sheen_sample(sc, Ng, sd->wi, randu, randv, eval, wo, pdf);
label = bsdf_principled_sheen_sample(sc, Ng, sd->I, randu, randv, eval, omega_in, pdf);
*sampled_roughness = one_float2();
*eta = 1.0f;
break;
@@ -253,12 +274,12 @@ ccl_device_inline int bsdf_sample(KernelGlobals kg,
const float frequency_multiplier =
kernel_data_fetch(objects, sd->object).shadow_terminator_shading_offset;
if (frequency_multiplier > 1.0f) {
const float cosNO = dot(*wo, sc->N);
*eval *= shift_cos_in(cosNO, frequency_multiplier);
const float cosNI = dot(*omega_in, sc->N);
*eval *= shift_cos_in(cosNI, frequency_multiplier);
}
if (label & LABEL_DIFFUSE) {
if (!isequal(sc->N, sd->N)) {
*eval *= bump_shadowing_term(sd->N, sc->N, *wo);
*eval *= bump_shadowing_term(sd->N, sc->N, *omega_in);
}
}
}
@@ -405,7 +426,7 @@ ccl_device_inline void bsdf_roughness_eta(const KernelGlobals kg,
ccl_device_inline int bsdf_label(const KernelGlobals kg,
ccl_private const ShaderClosure *sc,
const float3 wo)
const float3 omega_in)
{
/* For curves use the smooth normal, particularly for ribbons the geometric
* normal gives too much darkening otherwise. */
@@ -461,8 +482,8 @@ ccl_device_inline int bsdf_label(const KernelGlobals kg,
}
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID:
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID:
label = (bsdf_is_transmission(sc, wo)) ? LABEL_TRANSMIT | LABEL_GLOSSY :
LABEL_REFLECT | LABEL_GLOSSY;
label = (bsdf_is_transmission(sc, omega_in)) ? LABEL_TRANSMIT | LABEL_GLOSSY :
LABEL_REFLECT | LABEL_GLOSSY;
break;
case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID:
label = LABEL_REFLECT | LABEL_GLOSSY;
@@ -483,7 +504,7 @@ ccl_device_inline int bsdf_label(const KernelGlobals kg,
label = LABEL_TRANSMIT | LABEL_GLOSSY;
break;
case CLOSURE_BSDF_HAIR_PRINCIPLED_ID:
if (bsdf_is_transmission(sc, wo))
if (bsdf_is_transmission(sc, omega_in))
label = LABEL_TRANSMIT | LABEL_GLOSSY;
else
label = LABEL_REFLECT | LABEL_GLOSSY;
@@ -522,83 +543,83 @@ ccl_device_inline
bsdf_eval(KernelGlobals kg,
ccl_private ShaderData *sd,
ccl_private const ShaderClosure *sc,
const float3 wo,
const float3 omega_in,
ccl_private float *pdf)
{
Spectrum eval = zero_spectrum();
switch (sc->type) {
case CLOSURE_BSDF_DIFFUSE_ID:
eval = bsdf_diffuse_eval(sc, sd->wi, wo, pdf);
eval = bsdf_diffuse_eval(sc, sd->I, omega_in, pdf);
break;
#if defined(__SVM__) || defined(__OSL__)
case CLOSURE_BSDF_OREN_NAYAR_ID:
eval = bsdf_oren_nayar_eval(sc, sd->wi, wo, pdf);
eval = bsdf_oren_nayar_eval(sc, sd->I, omega_in, pdf);
break;
# ifdef __OSL__
case CLOSURE_BSDF_PHONG_RAMP_ID:
eval = bsdf_phong_ramp_eval(sc, sd->wi, wo, pdf);
eval = bsdf_phong_ramp_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_DIFFUSE_RAMP_ID:
eval = bsdf_diffuse_ramp_eval(sc, sd->wi, wo, pdf);
eval = bsdf_diffuse_ramp_eval(sc, sd->I, omega_in, pdf);
break;
# endif
case CLOSURE_BSDF_TRANSLUCENT_ID:
eval = bsdf_translucent_eval(sc, sd->wi, wo, pdf);
eval = bsdf_translucent_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_REFLECTION_ID:
eval = bsdf_reflection_eval(sc, sd->wi, wo, pdf);
eval = bsdf_reflection_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_REFRACTION_ID:
eval = bsdf_refraction_eval(sc, sd->wi, wo, pdf);
eval = bsdf_refraction_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_TRANSPARENT_ID:
eval = bsdf_transparent_eval(sc, sd->wi, wo, pdf);
eval = bsdf_transparent_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_MICROFACET_GGX_ID:
case CLOSURE_BSDF_MICROFACET_GGX_FRESNEL_ID:
case CLOSURE_BSDF_MICROFACET_GGX_CLEARCOAT_ID:
case CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID:
eval = bsdf_microfacet_ggx_eval(sc, sd->N, sd->wi, wo, pdf);
eval = bsdf_microfacet_ggx_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID:
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_FRESNEL_ID:
eval = bsdf_microfacet_multi_ggx_eval(sc, sd->N, sd->wi, wo, pdf, &sd->lcg_state);
eval = bsdf_microfacet_multi_ggx_eval(sc, sd->I, omega_in, pdf, &sd->lcg_state);
break;
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID:
case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID:
eval = bsdf_microfacet_multi_ggx_glass_eval(sc, sd->wi, wo, pdf, &sd->lcg_state);
eval = bsdf_microfacet_multi_ggx_glass_eval(sc, sd->I, omega_in, pdf, &sd->lcg_state);
break;
case CLOSURE_BSDF_MICROFACET_BECKMANN_ID:
case CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID:
eval = bsdf_microfacet_beckmann_eval(sc, sd->N, sd->wi, wo, pdf);
eval = bsdf_microfacet_beckmann_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID:
eval = bsdf_ashikhmin_shirley_eval(sc, sd->N, sd->wi, wo, pdf);
eval = bsdf_ashikhmin_shirley_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID:
eval = bsdf_ashikhmin_velvet_eval(sc, sd->wi, wo, pdf);
eval = bsdf_ashikhmin_velvet_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_DIFFUSE_TOON_ID:
eval = bsdf_diffuse_toon_eval(sc, sd->wi, wo, pdf);
eval = bsdf_diffuse_toon_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_GLOSSY_TOON_ID:
eval = bsdf_glossy_toon_eval(sc, sd->wi, wo, pdf);
eval = bsdf_glossy_toon_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_HAIR_PRINCIPLED_ID:
eval = bsdf_principled_hair_eval(kg, sd, sc, wo, pdf);
eval = bsdf_principled_hair_eval(kg, sd, sc, omega_in, pdf);
break;
case CLOSURE_BSDF_HAIR_REFLECTION_ID:
eval = bsdf_hair_reflection_eval(sc, sd->wi, wo, pdf);
eval = bsdf_hair_reflection_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_HAIR_TRANSMISSION_ID:
eval = bsdf_hair_transmission_eval(sc, sd->wi, wo, pdf);
eval = bsdf_hair_transmission_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_PRINCIPLED_DIFFUSE_ID:
eval = bsdf_principled_diffuse_eval(sc, sd->wi, wo, pdf);
eval = bsdf_principled_diffuse_eval(sc, sd->I, omega_in, pdf);
break;
case CLOSURE_BSDF_PRINCIPLED_SHEEN_ID:
eval = bsdf_principled_sheen_eval(sc, sd->wi, wo, pdf);
eval = bsdf_principled_sheen_eval(sc, sd->I, omega_in, pdf);
break;
#endif
default:
@@ -607,7 +628,7 @@ ccl_device_inline
if (CLOSURE_IS_BSDF_DIFFUSE(sc->type)) {
if (!isequal(sc->N, sd->N)) {
eval *= bump_shadowing_term(sd->N, sc->N, wo);
eval *= bump_shadowing_term(sd->N, sc->N, omega_in);
}
}
@@ -615,9 +636,9 @@ ccl_device_inline
const float frequency_multiplier =
kernel_data_fetch(objects, sd->object).shadow_terminator_shading_offset;
if (frequency_multiplier > 1.0f) {
const float cosNO = dot(wo, sc->N);
if (cosNO >= 0.0f) {
eval *= shift_cos_in(cosNO, frequency_multiplier);
const float cosNI = dot(omega_in, sc->N);
if (cosNI >= 0.0f) {
eval *= shift_cos_in(cosNI, frequency_multiplier);
}
}

View File

@@ -40,37 +40,34 @@ ccl_device_inline float bsdf_ashikhmin_shirley_roughness_to_exponent(float rough
}
ccl_device_forceinline Spectrum bsdf_ashikhmin_shirley_eval(ccl_private const ShaderClosure *sc,
const float3 Ng,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const MicrofacetBsdf *bsdf = (ccl_private const MicrofacetBsdf *)sc;
const float cosNgO = dot(Ng, wo);
float3 N = bsdf->N;
float NdotI = dot(N, wi);
float NdotO = dot(N, wo);
float NdotI = dot(N, I); /* in Cycles/OSL convention I is omega_out */
float NdotO = dot(N, omega_in); /* and consequently we use for O omaga_in ;) */
float out = 0.0f;
if ((cosNgO < 0.0f) || fmaxf(bsdf->alpha_x, bsdf->alpha_y) <= 1e-4f ||
!(NdotI > 0.0f && NdotO > 0.0f)) {
if (fmaxf(bsdf->alpha_x, bsdf->alpha_y) <= 1e-4f || !(NdotI > 0.0f && NdotO > 0.0f)) {
*pdf = 0.0f;
return zero_spectrum();
}
NdotI = fmaxf(NdotI, 1e-6f);
NdotO = fmaxf(NdotO, 1e-6f);
float3 H = normalize(wi + wo);
float HdotI = fmaxf(fabsf(dot(H, wi)), 1e-6f);
float3 H = normalize(omega_in + I);
float HdotI = fmaxf(fabsf(dot(H, I)), 1e-6f);
float HdotN = fmaxf(dot(H, N), 1e-6f);
/* pump from original paper
* (first derivative disc., but cancels the HdotI in the pdf nicely) */
float pump = 1.0f / fmaxf(1e-6f, (HdotI * fmaxf(NdotI, NdotO)));
float pump = 1.0f / fmaxf(1e-6f, (HdotI * fmaxf(NdotO, NdotI)));
/* pump from d-brdf paper */
/*float pump = 1.0f / fmaxf(1e-4f, ((NdotI + NdotO) * (NdotI * NdotO))); */
/*float pump = 1.0f / fmaxf(1e-4f, ((NdotO + NdotI) * (NdotO*NdotI))); */
float n_x = bsdf_ashikhmin_shirley_roughness_to_exponent(bsdf->alpha_x);
float n_y = bsdf_ashikhmin_shirley_roughness_to_exponent(bsdf->alpha_y);
@@ -124,11 +121,11 @@ ccl_device_inline void bsdf_ashikhmin_shirley_sample_first_quadrant(float n_x,
ccl_device int bsdf_ashikhmin_shirley_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness)
{
@@ -137,7 +134,7 @@ ccl_device int bsdf_ashikhmin_shirley_sample(ccl_private const ShaderClosure *sc
float3 N = bsdf->N;
int label = LABEL_REFLECT | LABEL_GLOSSY;
float NdotI = dot(N, wi);
float NdotI = dot(N, I);
if (!(NdotI > 0.0f)) {
*pdf = 0.0f;
*eval = zero_spectrum();
@@ -198,12 +195,12 @@ ccl_device int bsdf_ashikhmin_shirley_sample(ccl_private const ShaderClosure *sc
/* half vector to world space */
float3 H = h.x * X + h.y * Y + h.z * N;
float HdotI = dot(H, wi);
float HdotI = dot(H, I);
if (HdotI < 0.0f)
H = -H;
/* reflect wi on H to get wo */
*wo = -wi + (2.0f * HdotI) * H;
/* reflect I on H to get omega_in */
*omega_in = -I + (2.0f * HdotI) * H;
if (fmaxf(bsdf->alpha_x, bsdf->alpha_y) <= 1e-4f) {
/* Some high number for MIS. */
@@ -213,7 +210,7 @@ ccl_device int bsdf_ashikhmin_shirley_sample(ccl_private const ShaderClosure *sc
}
else {
/* leave the rest to eval */
*eval = bsdf_ashikhmin_shirley_eval(sc, N, wi, *wo, pdf);
*eval = bsdf_ashikhmin_shirley_eval(sc, I, *omega_in, pdf);
}
return label;

View File

@@ -32,35 +32,35 @@ ccl_device int bsdf_ashikhmin_velvet_setup(ccl_private VelvetBsdf *bsdf)
}
ccl_device Spectrum bsdf_ashikhmin_velvet_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const VelvetBsdf *bsdf = (ccl_private const VelvetBsdf *)sc;
float m_invsigma2 = bsdf->invsigma2;
float3 N = bsdf->N;
float cosNI = dot(N, wi);
float cosNO = dot(N, wo);
if (!(cosNI > 0 && cosNO > 0)) {
float cosNO = dot(N, I);
float cosNI = dot(N, omega_in);
if (!(cosNO > 0 && cosNI > 0)) {
*pdf = 0.0f;
return zero_spectrum();
}
float3 H = normalize(wi + wo);
float3 H = normalize(omega_in + I);
float cosNH = dot(N, H);
float cosHI = fabsf(dot(wi, H));
float cosHO = fabsf(dot(I, H));
if (!(fabsf(cosNH) < 1.0f - 1e-5f && cosHI > 1e-5f)) {
if (!(fabsf(cosNH) < 1.0f - 1e-5f && cosHO > 1e-5f)) {
*pdf = 0.0f;
return zero_spectrum();
}
float cosNHdivHI = cosNH / cosHI;
cosNHdivHI = fmaxf(cosNHdivHI, 1e-5f);
float cosNHdivHO = cosNH / cosHO;
cosNHdivHO = fmaxf(cosNHdivHO, 1e-5f);
float fac1 = 2 * fabsf(cosNHdivHI * cosNI);
float fac2 = 2 * fabsf(cosNHdivHI * cosNO);
float fac1 = 2 * fabsf(cosNHdivHO * cosNO);
float fac2 = 2 * fabsf(cosNHdivHO * cosNI);
float sinNH2 = 1 - cosNH * cosNH;
float sinNH4 = sinNH2 * sinNH2;
@@ -69,7 +69,7 @@ ccl_device Spectrum bsdf_ashikhmin_velvet_eval(ccl_private const ShaderClosure *
float D = expf(-cotangent2 * m_invsigma2) * m_invsigma2 * M_1_PI_F / sinNH4;
float G = fminf(1.0f, fminf(fac1, fac2)); // TODO: derive G from D analytically
float out = 0.25f * (D * G) / cosNI;
float out = 0.25f * (D * G) / cosNO;
*pdf = 0.5f * M_1_PI_F;
return make_spectrum(out);
@@ -77,11 +77,11 @@ ccl_device Spectrum bsdf_ashikhmin_velvet_eval(ccl_private const ShaderClosure *
ccl_device int bsdf_ashikhmin_velvet_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf)
{
ccl_private const VelvetBsdf *bsdf = (ccl_private const VelvetBsdf *)sc;
@@ -90,32 +90,32 @@ ccl_device int bsdf_ashikhmin_velvet_sample(ccl_private const ShaderClosure *sc,
// we are viewing the surface from above - send a ray out with uniform
// distribution over the hemisphere
sample_uniform_hemisphere(N, randu, randv, wo, pdf);
sample_uniform_hemisphere(N, randu, randv, omega_in, pdf);
if (!(dot(Ng, *wo) > 0)) {
if (!(dot(Ng, *omega_in) > 0)) {
*pdf = 0.0f;
*eval = zero_spectrum();
return LABEL_NONE;
}
float3 H = normalize(wi + *wo);
float3 H = normalize(*omega_in + I);
float cosNI = dot(N, wi);
float cosNO = dot(N, *wo);
float cosHI = fabsf(dot(wi, H));
float cosNI = dot(N, *omega_in);
float cosNO = dot(N, I);
float cosNH = dot(N, H);
float cosHO = fabsf(dot(I, H));
if (!(fabsf(cosNI) > 1e-5f && fabsf(cosNH) < 1.0f - 1e-5f && cosHI > 1e-5f)) {
if (!(fabsf(cosNO) > 1e-5f && fabsf(cosNH) < 1.0f - 1e-5f && cosHO > 1e-5f)) {
*pdf = 0.0f;
*eval = zero_spectrum();
return LABEL_NONE;
}
float cosNHdivHI = cosNH / cosHI;
cosNHdivHI = fmaxf(cosNHdivHI, 1e-5f);
float cosNHdivHO = cosNH / cosHO;
cosNHdivHO = fmaxf(cosNHdivHO, 1e-5f);
float fac1 = 2 * fabsf(cosNHdivHI * cosNI);
float fac2 = 2 * fabsf(cosNHdivHI * cosNO);
float fac1 = 2 * fabsf(cosNHdivHO * cosNO);
float fac2 = 2 * fabsf(cosNHdivHO * cosNI);
float sinNH2 = 1 - cosNH * cosNH;
float sinNH4 = sinNH2 * sinNH2;
@@ -124,7 +124,7 @@ ccl_device int bsdf_ashikhmin_velvet_sample(ccl_private const ShaderClosure *sc,
float D = expf(-cotangent2 * m_invsigma2) * m_invsigma2 * M_1_PI_F / sinNH4;
float G = fminf(1.0f, fminf(fac1, fac2)); // TODO: derive G from D analytically
float power = 0.25f * (D * G) / cosNI;
float power = 0.25f * (D * G) / cosNO;
*eval = make_spectrum(power);

View File

@@ -27,34 +27,34 @@ ccl_device int bsdf_diffuse_setup(ccl_private DiffuseBsdf *bsdf)
}
ccl_device Spectrum bsdf_diffuse_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const DiffuseBsdf *bsdf = (ccl_private const DiffuseBsdf *)sc;
float3 N = bsdf->N;
float cosNO = fmaxf(dot(N, wo), 0.0f) * M_1_PI_F;
*pdf = cosNO;
return make_spectrum(cosNO);
float cos_pi = fmaxf(dot(N, omega_in), 0.0f) * M_1_PI_F;
*pdf = cos_pi;
return make_spectrum(cos_pi);
}
ccl_device int bsdf_diffuse_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf)
{
ccl_private const DiffuseBsdf *bsdf = (ccl_private const DiffuseBsdf *)sc;
float3 N = bsdf->N;
// distribution over the hemisphere
sample_cos_hemisphere(N, randu, randv, wo, pdf);
sample_cos_hemisphere(N, randu, randv, omega_in, pdf);
if (dot(Ng, *wo) > 0.0f) {
if (dot(Ng, *omega_in) > 0.0f) {
*eval = make_spectrum(*pdf);
}
else {
@@ -73,25 +73,25 @@ ccl_device int bsdf_translucent_setup(ccl_private DiffuseBsdf *bsdf)
}
ccl_device Spectrum bsdf_translucent_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const DiffuseBsdf *bsdf = (ccl_private const DiffuseBsdf *)sc;
float3 N = bsdf->N;
float cosNO = fmaxf(-dot(N, wo), 0.0f) * M_1_PI_F;
*pdf = cosNO;
return make_spectrum(cosNO);
float cos_pi = fmaxf(-dot(N, omega_in), 0.0f) * M_1_PI_F;
*pdf = cos_pi;
return make_spectrum(cos_pi);
}
ccl_device int bsdf_translucent_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf)
{
ccl_private const DiffuseBsdf *bsdf = (ccl_private const DiffuseBsdf *)sc;
@@ -99,8 +99,8 @@ ccl_device int bsdf_translucent_sample(ccl_private const ShaderClosure *sc,
// we are viewing the surface from the right side - send a ray out with cosine
// distribution over the hemisphere
sample_cos_hemisphere(-N, randu, randv, wo, pdf);
if (dot(Ng, *wo) < 0) {
sample_cos_hemisphere(-N, randu, randv, omega_in, pdf);
if (dot(Ng, *omega_in) < 0) {
*eval = make_spectrum(*pdf);
}
else {

View File

@@ -48,17 +48,17 @@ ccl_device void bsdf_diffuse_ramp_blur(ccl_private ShaderClosure *sc, float roug
}
ccl_device Spectrum bsdf_diffuse_ramp_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
const DiffuseRampBsdf *bsdf = (const DiffuseRampBsdf *)sc;
float3 N = bsdf->N;
float cosNO = fmaxf(dot(N, wo), 0.0f);
if (cosNO >= 0.0f) {
*pdf = cosNO * M_1_PI_F;
return rgb_to_spectrum(bsdf_diffuse_ramp_get_color(bsdf->colors, cosNO) * M_1_PI_F);
float cos_pi = fmaxf(dot(N, omega_in), 0.0f);
if (cos_pi >= 0.0f) {
*pdf = cos_pi * M_1_PI_F;
return rgb_to_spectrum(bsdf_diffuse_ramp_get_color(bsdf->colors, cos_pi) * M_1_PI_F);
}
else {
*pdf = 0.0f;
@@ -68,20 +68,20 @@ ccl_device Spectrum bsdf_diffuse_ramp_eval(ccl_private const ShaderClosure *sc,
ccl_device int bsdf_diffuse_ramp_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf)
{
const DiffuseRampBsdf *bsdf = (const DiffuseRampBsdf *)sc;
float3 N = bsdf->N;
// distribution over the hemisphere
sample_cos_hemisphere(N, randu, randv, wo, pdf);
sample_cos_hemisphere(N, randu, randv, omega_in, pdf);
if (dot(Ng, *wo) > 0.0f) {
if (dot(Ng, *omega_in) > 0.0f) {
*eval = rgb_to_spectrum(bsdf_diffuse_ramp_get_color(bsdf->colors, *pdf * M_PI_F) * M_1_PI_F);
}
else {

View File

@@ -38,12 +38,12 @@ ccl_device int bsdf_hair_transmission_setup(ccl_private HairBsdf *bsdf)
}
ccl_device Spectrum bsdf_hair_reflection_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const HairBsdf *bsdf = (ccl_private const HairBsdf *)sc;
if (dot(bsdf->N, wo) < 0.0f) {
if (dot(bsdf->N, omega_in) < 0.0f) {
*pdf = 0.0f;
return zero_spectrum();
}
@@ -53,16 +53,16 @@ ccl_device Spectrum bsdf_hair_reflection_eval(ccl_private const ShaderClosure *s
float roughness1 = bsdf->roughness1;
float roughness2 = bsdf->roughness2;
float Iz = dot(Tg, wi);
float3 locy = normalize(wi - Tg * Iz);
float Iz = dot(Tg, I);
float3 locy = normalize(I - Tg * Iz);
float theta_r = M_PI_2_F - fast_acosf(Iz);
float wo_z = dot(Tg, wo);
float3 wo_y = normalize(wo - Tg * wo_z);
float omega_in_z = dot(Tg, omega_in);
float3 omega_in_y = normalize(omega_in - Tg * omega_in_z);
float theta_i = M_PI_2_F - fast_acosf(wo_z);
float cosphi_i = dot(wo_y, locy);
float theta_i = M_PI_2_F - fast_acosf(omega_in_z);
float cosphi_i = dot(omega_in_y, locy);
if (M_PI_2_F - fabsf(theta_i) < 0.001f || cosphi_i < 0.0f) {
*pdf = 0.0f;
@@ -90,12 +90,12 @@ ccl_device Spectrum bsdf_hair_reflection_eval(ccl_private const ShaderClosure *s
}
ccl_device Spectrum bsdf_hair_transmission_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const HairBsdf *bsdf = (ccl_private const HairBsdf *)sc;
if (dot(bsdf->N, wo) >= 0.0f) {
if (dot(bsdf->N, omega_in) >= 0.0f) {
*pdf = 0.0f;
return zero_spectrum();
}
@@ -104,16 +104,16 @@ ccl_device Spectrum bsdf_hair_transmission_eval(ccl_private const ShaderClosure
float3 Tg = bsdf->T;
float roughness1 = bsdf->roughness1;
float roughness2 = bsdf->roughness2;
float Iz = dot(Tg, wi);
float3 locy = normalize(wi - Tg * Iz);
float Iz = dot(Tg, I);
float3 locy = normalize(I - Tg * Iz);
float theta_r = M_PI_2_F - fast_acosf(Iz);
float wo_z = dot(Tg, wo);
float3 wo_y = normalize(wo - Tg * wo_z);
float omega_in_z = dot(Tg, omega_in);
float3 omega_in_y = normalize(omega_in - Tg * omega_in_z);
float theta_i = M_PI_2_F - fast_acosf(wo_z);
float phi_i = fast_acosf(dot(wo_y, locy));
float theta_i = M_PI_2_F - fast_acosf(omega_in_z);
float phi_i = fast_acosf(dot(omega_in_y, locy));
if (M_PI_2_F - fabsf(theta_i) < 0.001f) {
*pdf = 0.0f;
@@ -142,11 +142,11 @@ ccl_device Spectrum bsdf_hair_transmission_eval(ccl_private const ShaderClosure
ccl_device int bsdf_hair_reflection_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness)
{
@@ -156,8 +156,8 @@ ccl_device int bsdf_hair_reflection_sample(ccl_private const ShaderClosure *sc,
float roughness1 = bsdf->roughness1;
float roughness2 = bsdf->roughness2;
*sampled_roughness = make_float2(roughness1, roughness2);
float Iz = dot(Tg, wi);
float3 locy = normalize(wi - Tg * Iz);
float Iz = dot(Tg, I);
float3 locy = normalize(I - Tg * Iz);
float3 locx = cross(locy, Tg);
float theta_r = M_PI_2_F - fast_acosf(Iz);
@@ -182,7 +182,7 @@ ccl_device int bsdf_hair_reflection_sample(ccl_private const ShaderClosure *sc,
float sinphi, cosphi;
fast_sincosf(phi, &sinphi, &cosphi);
*wo = (cosphi * costheta_i) * locy - (sinphi * costheta_i) * locx + (sintheta_i)*Tg;
*omega_in = (cosphi * costheta_i) * locy - (sinphi * costheta_i) * locx + (sintheta_i)*Tg;
*pdf = fabsf(phi_pdf * theta_pdf);
if (M_PI_2_F - fabsf(theta_i) < 0.001f)
@@ -195,11 +195,11 @@ ccl_device int bsdf_hair_reflection_sample(ccl_private const ShaderClosure *sc,
ccl_device int bsdf_hair_transmission_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness)
{
@@ -209,8 +209,8 @@ ccl_device int bsdf_hair_transmission_sample(ccl_private const ShaderClosure *sc
float roughness1 = bsdf->roughness1;
float roughness2 = bsdf->roughness2;
*sampled_roughness = make_float2(roughness1, roughness2);
float Iz = dot(Tg, wi);
float3 locy = normalize(wi - Tg * Iz);
float Iz = dot(Tg, I);
float3 locy = normalize(I - Tg * Iz);
float3 locx = cross(locy, Tg);
float theta_r = M_PI_2_F - fast_acosf(Iz);
@@ -235,7 +235,7 @@ ccl_device int bsdf_hair_transmission_sample(ccl_private const ShaderClosure *sc
float sinphi, cosphi;
fast_sincosf(phi, &sinphi, &cosphi);
*wo = (cosphi * costheta_i) * locy - (sinphi * costheta_i) * locx + (sintheta_i)*Tg;
*omega_in = (cosphi * costheta_i) * locy - (sinphi * costheta_i) * locx + (sintheta_i)*Tg;
*pdf = fabsf(phi_pdf * theta_pdf);
if (M_PI_2_F - fabsf(theta_i) < 0.001f) {
@@ -247,7 +247,7 @@ ccl_device int bsdf_hair_transmission_sample(ccl_private const ShaderClosure *sc
/* TODO(sergey): Should always be negative, but seems some precision issue
* is involved here.
*/
kernel_assert(dot(locy, *wo) < 1e-4f);
kernel_assert(dot(locy, *omega_in) < 1e-4f);
return LABEL_TRANSMIT | LABEL_GLOSSY;
}

View File

@@ -179,7 +179,7 @@ ccl_device int bsdf_principled_hair_setup(ccl_private ShaderData *sd,
/* Compute local frame, aligned to curve tangent and ray direction. */
float3 X = safe_normalize(sd->dPdu);
float3 Y = safe_normalize(cross(X, sd->wi));
float3 Y = safe_normalize(cross(X, sd->I));
float3 Z = safe_normalize(cross(X, Y));
/* h -1..0..1 means the rays goes from grazing the hair, to hitting it at
@@ -259,7 +259,7 @@ ccl_device_inline void hair_alpha_angles(float sin_theta_i,
ccl_device Spectrum bsdf_principled_hair_eval(KernelGlobals kg,
ccl_private const ShaderData *sd,
ccl_private const ShaderClosure *sc,
const float3 wo,
const float3 omega_in,
ccl_private float *pdf)
{
kernel_assert(isfinite_safe(sd->P) && isfinite_safe(sd->ray_length));
@@ -271,13 +271,12 @@ ccl_device Spectrum bsdf_principled_hair_eval(KernelGlobals kg,
kernel_assert(fabsf(dot(X, Y)) < 1e-3f);
const float3 Z = safe_normalize(cross(X, Y));
/* local_I is the illumination direction. */
const float3 local_O = make_float3(dot(sd->wi, X), dot(sd->wi, Y), dot(sd->wi, Z));
const float3 local_I = make_float3(dot(wo, X), dot(wo, Y), dot(wo, Z));
const float3 wo = make_float3(dot(sd->I, X), dot(sd->I, Y), dot(sd->I, Z));
const float3 wi = make_float3(dot(omega_in, X), dot(omega_in, Y), dot(omega_in, Z));
const float sin_theta_o = local_O.x;
const float sin_theta_o = wo.x;
const float cos_theta_o = cos_from_sin(sin_theta_o);
const float phi_o = atan2f(local_O.z, local_O.y);
const float phi_o = atan2f(wo.z, wo.y);
const float sin_theta_t = sin_theta_o / bsdf->eta;
const float cos_theta_t = cos_from_sin(sin_theta_t);
@@ -296,9 +295,9 @@ ccl_device Spectrum bsdf_principled_hair_eval(KernelGlobals kg,
hair_attenuation(
kg, fresnel_dielectric_cos(cos_theta_o * cos_gamma_o, bsdf->eta), T, Ap, Ap_energy);
const float sin_theta_i = local_I.x;
const float sin_theta_i = wi.x;
const float cos_theta_i = cos_from_sin(sin_theta_i);
const float phi_i = atan2f(local_I.z, local_I.y);
const float phi_i = atan2f(wi.z, wi.y);
const float phi = phi_i - phi_o;
@@ -344,7 +343,7 @@ ccl_device int bsdf_principled_hair_sample(KernelGlobals kg,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness,
ccl_private float *eta)
@@ -360,16 +359,16 @@ ccl_device int bsdf_principled_hair_sample(KernelGlobals kg,
kernel_assert(fabsf(dot(X, Y)) < 1e-3f);
const float3 Z = safe_normalize(cross(X, Y));
const float3 local_O = make_float3(dot(sd->wi, X), dot(sd->wi, Y), dot(sd->wi, Z));
const float3 wo = make_float3(dot(sd->I, X), dot(sd->I, Y), dot(sd->I, Z));
float2 u[2];
u[0] = make_float2(randu, randv);
u[1].x = lcg_step_float(&sd->lcg_state);
u[1].y = lcg_step_float(&sd->lcg_state);
const float sin_theta_o = local_O.x;
const float sin_theta_o = wo.x;
const float cos_theta_o = cos_from_sin(sin_theta_o);
const float phi_o = atan2f(local_O.z, local_O.y);
const float phi_o = atan2f(wo.z, wo.y);
const float sin_theta_t = sin_theta_o / bsdf->eta;
const float cos_theta_t = cos_from_sin(sin_theta_t);
@@ -459,7 +458,7 @@ ccl_device int bsdf_principled_hair_sample(KernelGlobals kg,
*eval = F;
*pdf = F_energy;
*wo = X * sin_theta_i + Y * cos_theta_i * cosf(phi_i) + Z * cos_theta_i * sinf(phi_i);
*omega_in = X * sin_theta_i + Y * cos_theta_i * cosf(phi_i) + Z * cos_theta_i * sinf(phi_i);
return LABEL_GLOSSY | ((p == 0) ? LABEL_REFLECT : LABEL_TRANSMIT);
}

File diff suppressed because it is too large Load Diff

View File

@@ -416,16 +416,14 @@ ccl_device int bsdf_microfacet_multi_ggx_refraction_setup(ccl_private Microfacet
}
ccl_device Spectrum bsdf_microfacet_multi_ggx_eval(ccl_private const ShaderClosure *sc,
const float3 Ng,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf,
ccl_private uint *lcg_state)
{
ccl_private const MicrofacetBsdf *bsdf = (ccl_private const MicrofacetBsdf *)sc;
const float cosNgO = dot(Ng, wo);
if ((cosNgO < 0.0f) || bsdf->alpha_x * bsdf->alpha_y < 1e-7f) {
if (bsdf->alpha_x * bsdf->alpha_y < 1e-7f) {
*pdf = 0.0f;
return zero_spectrum();
}
@@ -434,7 +432,7 @@ ccl_device Spectrum bsdf_microfacet_multi_ggx_eval(ccl_private const ShaderClosu
Z = bsdf->N;
/* Ensure that the both directions are on the outside w.r.t. the shading normal. */
if (dot(Z, wi) <= 0.0f || dot(Z, wo) <= 0.0f) {
if (dot(Z, I) <= 0.0f || dot(Z, omega_in) <= 0.0f) {
*pdf = 0.0f;
return zero_spectrum();
}
@@ -447,21 +445,21 @@ ccl_device Spectrum bsdf_microfacet_multi_ggx_eval(ccl_private const ShaderClosu
else
make_orthonormals(Z, &X, &Y);
float3 local_I = make_float3(dot(wi, X), dot(wi, Y), dot(wi, Z));
float3 local_O = make_float3(dot(wo, X), dot(wo, Y), dot(wo, Z));
float3 localI = make_float3(dot(I, X), dot(I, Y), dot(I, Z));
float3 localO = make_float3(dot(omega_in, X), dot(omega_in, Y), dot(omega_in, Z));
if (is_aniso)
*pdf = mf_ggx_aniso_pdf(local_I, local_O, make_float2(bsdf->alpha_x, bsdf->alpha_y));
*pdf = mf_ggx_aniso_pdf(localI, localO, make_float2(bsdf->alpha_x, bsdf->alpha_y));
else
*pdf = mf_ggx_pdf(local_I, local_O, bsdf->alpha_x);
*pdf = mf_ggx_pdf(localI, localO, bsdf->alpha_x);
if (*pdf <= 0.f) {
*pdf = 0.f;
return make_float3(0.f, 0.f, 0.f);
}
return mf_eval_glossy(local_I,
local_O,
return mf_eval_glossy(localI,
localO,
true,
bsdf->extra->color,
bsdf->alpha_x,
@@ -475,11 +473,11 @@ ccl_device Spectrum bsdf_microfacet_multi_ggx_eval(ccl_private const ShaderClosu
ccl_device int bsdf_microfacet_multi_ggx_sample(KernelGlobals kg,
ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private uint *lcg_state,
ccl_private float2 *sampled_roughness,
@@ -491,7 +489,7 @@ ccl_device int bsdf_microfacet_multi_ggx_sample(KernelGlobals kg,
Z = bsdf->N;
/* Ensure that the view direction is on the outside w.r.t. the shading normal. */
if (dot(Z, wi) <= 0.0f) {
if (dot(Z, I) <= 0.0f) {
*pdf = 0.0f;
return LABEL_NONE;
}
@@ -499,8 +497,8 @@ ccl_device int bsdf_microfacet_multi_ggx_sample(KernelGlobals kg,
/* Special case: Extremely low roughness.
* Don't bother with microfacets, just do specular reflection. */
if (bsdf->alpha_x * bsdf->alpha_y < 1e-7f) {
*wo = 2 * dot(Z, wi) * Z - wi;
if (dot(Ng, *wo) <= 0.0f) {
*omega_in = 2 * dot(Z, I) * Z - I;
if (dot(Ng, *omega_in) <= 0.0f) {
*pdf = 0.0f;
return LABEL_NONE;
}
@@ -520,11 +518,11 @@ ccl_device int bsdf_microfacet_multi_ggx_sample(KernelGlobals kg,
else
make_orthonormals(Z, &X, &Y);
float3 local_I = make_float3(dot(wi, X), dot(wi, Y), dot(wi, Z));
float3 local_O;
float3 localI = make_float3(dot(I, X), dot(I, Y), dot(I, Z));
float3 localO;
*eval = mf_sample_glossy(local_I,
&local_O,
*eval = mf_sample_glossy(localI,
&localO,
bsdf->extra->color,
bsdf->alpha_x,
bsdf->alpha_y,
@@ -532,18 +530,18 @@ ccl_device int bsdf_microfacet_multi_ggx_sample(KernelGlobals kg,
bsdf->ior,
use_fresnel,
bsdf->extra->cspec0);
*wo = X * local_O.x + Y * local_O.y + Z * local_O.z;
*omega_in = X * localO.x + Y * localO.y + Z * localO.z;
/* Ensure that the light direction is on the outside w.r.t. the geometry normal. */
if (dot(Ng, *wo) <= 0.0f) {
if (dot(Ng, *omega_in) <= 0.0f) {
*pdf = 0.0f;
return LABEL_NONE;
}
if (is_aniso)
*pdf = mf_ggx_aniso_pdf(local_I, local_O, make_float2(bsdf->alpha_x, bsdf->alpha_y));
*pdf = mf_ggx_aniso_pdf(localI, localO, make_float2(bsdf->alpha_x, bsdf->alpha_y));
else
*pdf = mf_ggx_pdf(local_I, local_O, bsdf->alpha_x);
*pdf = mf_ggx_pdf(localI, localO, bsdf->alpha_x);
*pdf = fmaxf(0.f, *pdf);
*eval *= *pdf;
@@ -581,8 +579,8 @@ ccl_device int bsdf_microfacet_multi_ggx_glass_fresnel_setup(ccl_private Microfa
}
ccl_device Spectrum bsdf_microfacet_multi_ggx_glass_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf,
ccl_private uint *lcg_state)
{
@@ -597,17 +595,17 @@ ccl_device Spectrum bsdf_microfacet_multi_ggx_glass_eval(ccl_private const Shade
Z = bsdf->N;
make_orthonormals(Z, &X, &Y);
float3 local_I = make_float3(dot(wi, X), dot(wi, Y), dot(wi, Z));
float3 local_O = make_float3(dot(wo, X), dot(wo, Y), dot(wo, Z));
float3 localI = make_float3(dot(I, X), dot(I, Y), dot(I, Z));
float3 localO = make_float3(dot(omega_in, X), dot(omega_in, Y), dot(omega_in, Z));
const bool is_transmission = local_O.z < 0.0f;
const bool is_transmission = localO.z < 0.0f;
const bool use_fresnel = !is_transmission &&
(bsdf->type == CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID);
*pdf = mf_glass_pdf(local_I, local_O, bsdf->alpha_x, bsdf->ior);
*pdf = mf_glass_pdf(localI, localO, bsdf->alpha_x, bsdf->ior);
kernel_assert(*pdf >= 0.f);
return mf_eval_glass(local_I,
local_O,
return mf_eval_glass(localI,
localO,
!is_transmission,
bsdf->extra->color,
bsdf->alpha_x,
@@ -621,11 +619,11 @@ ccl_device Spectrum bsdf_microfacet_multi_ggx_glass_eval(ccl_private const Shade
ccl_device int bsdf_microfacet_multi_ggx_glass_sample(KernelGlobals kg,
ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private uint *lcg_state,
ccl_private float2 *sampled_roughness,
@@ -642,16 +640,16 @@ ccl_device int bsdf_microfacet_multi_ggx_glass_sample(KernelGlobals kg,
if (bsdf->alpha_x * bsdf->alpha_y < 1e-7f) {
float3 R, T;
bool inside;
float fresnel = fresnel_dielectric(bsdf->ior, Z, wi, &R, &T, &inside);
float fresnel = fresnel_dielectric(bsdf->ior, Z, I, &R, &T, &inside);
*pdf = 1e6f;
*eval = make_spectrum(1e6f);
if (randu < fresnel) {
*wo = R;
*omega_in = R;
return LABEL_REFLECT | LABEL_SINGULAR;
}
else {
*wo = T;
*omega_in = T;
return LABEL_TRANSMIT | LABEL_SINGULAR;
}
}
@@ -660,11 +658,11 @@ ccl_device int bsdf_microfacet_multi_ggx_glass_sample(KernelGlobals kg,
make_orthonormals(Z, &X, &Y);
float3 local_I = make_float3(dot(wi, X), dot(wi, Y), dot(wi, Z));
float3 local_O;
float3 localI = make_float3(dot(I, X), dot(I, Y), dot(I, Z));
float3 localO;
*eval = mf_sample_glass(local_I,
&local_O,
*eval = mf_sample_glass(localI,
&localO,
bsdf->extra->color,
bsdf->alpha_x,
bsdf->alpha_y,
@@ -672,12 +670,12 @@ ccl_device int bsdf_microfacet_multi_ggx_glass_sample(KernelGlobals kg,
bsdf->ior,
use_fresnel,
bsdf->extra->cspec0);
*pdf = mf_glass_pdf(local_I, local_O, bsdf->alpha_x, bsdf->ior);
*pdf = mf_glass_pdf(localI, localO, bsdf->alpha_x, bsdf->ior);
kernel_assert(*pdf >= 0.f);
*eval *= *pdf;
*wo = X * local_O.x + Y * local_O.y + Z * local_O.z;
if (local_O.z * local_I.z > 0.0f) {
*omega_in = X * localO.x + Y * localO.y + Z * localO.z;
if (localO.z * localI.z > 0.0f) {
return LABEL_REFLECT | LABEL_GLOSSY;
}
else {

View File

@@ -48,14 +48,14 @@ ccl_device int bsdf_oren_nayar_setup(ccl_private OrenNayarBsdf *bsdf)
}
ccl_device Spectrum bsdf_oren_nayar_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const OrenNayarBsdf *bsdf = (ccl_private const OrenNayarBsdf *)sc;
if (dot(bsdf->N, wo) > 0.0f) {
if (dot(bsdf->N, omega_in) > 0.0f) {
*pdf = 0.5f * M_1_PI_F;
return bsdf_oren_nayar_get_intensity(sc, bsdf->N, wi, wo);
return bsdf_oren_nayar_get_intensity(sc, bsdf->N, I, omega_in);
}
else {
*pdf = 0.0f;
@@ -65,18 +65,18 @@ ccl_device Spectrum bsdf_oren_nayar_eval(ccl_private const ShaderClosure *sc,
ccl_device int bsdf_oren_nayar_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf)
{
ccl_private const OrenNayarBsdf *bsdf = (ccl_private const OrenNayarBsdf *)sc;
sample_uniform_hemisphere(bsdf->N, randu, randv, wo, pdf);
sample_uniform_hemisphere(bsdf->N, randu, randv, omega_in, pdf);
if (dot(Ng, *wo) > 0.0f) {
*eval = bsdf_oren_nayar_get_intensity(sc, bsdf->N, wi, *wo);
if (dot(Ng, *omega_in) > 0.0f) {
*eval = bsdf_oren_nayar_get_intensity(sc, bsdf->N, I, *omega_in);
}
else {
*pdf = 0.0f;

View File

@@ -45,23 +45,23 @@ ccl_device int bsdf_phong_ramp_setup(ccl_private PhongRampBsdf *bsdf)
}
ccl_device Spectrum bsdf_phong_ramp_eval(ccl_private const ShaderClosure *sc,
const float3 wi,
const float3 wo,
const float3 I,
const float3 omega_in,
ccl_private float *pdf)
{
ccl_private const PhongRampBsdf *bsdf = (ccl_private const PhongRampBsdf *)sc;
float m_exponent = bsdf->exponent;
float cosNI = dot(bsdf->N, wi);
float cosNO = dot(bsdf->N, wo);
float cosNI = dot(bsdf->N, omega_in);
float cosNO = dot(bsdf->N, I);
if (cosNI > 0 && cosNO > 0) {
// reflect the view vector
float3 R = (2 * cosNI) * bsdf->N - wi;
float cosRO = dot(R, wo);
if (cosRO > 0) {
float cosp = powf(cosRO, m_exponent);
float3 R = (2 * cosNO) * bsdf->N - I;
float cosRI = dot(R, omega_in);
if (cosRI > 0) {
float cosp = powf(cosRI, m_exponent);
float common = 0.5f * M_1_PI_F * cosp;
float out = cosNO * (m_exponent + 2) * common;
float out = cosNI * (m_exponent + 2) * common;
*pdf = (m_exponent + 1) * common;
return rgb_to_spectrum(bsdf_phong_ramp_get_color(bsdf->colors, cosp) * out);
}
@@ -77,39 +77,39 @@ ccl_device_inline float phong_ramp_exponent_to_roughness(float exponent)
ccl_device int bsdf_phong_ramp_sample(ccl_private const ShaderClosure *sc,
float3 Ng,
float3 wi,
float3 I,
float randu,
float randv,
ccl_private Spectrum *eval,
ccl_private float3 *wo,
ccl_private float3 *omega_in,
ccl_private float *pdf,
ccl_private float2 *sampled_roughness)
{
ccl_private const PhongRampBsdf *bsdf = (ccl_private const PhongRampBsdf *)sc;
float cosNI = dot(bsdf->N, wi);
float cosNO = dot(bsdf->N, I);
float m_exponent = bsdf->exponent;
const float m_roughness = phong_ramp_exponent_to_roughness(m_exponent);
*sampled_roughness = make_float2(m_roughness, m_roughness);
if (cosNI > 0) {
if (cosNO > 0) {
// reflect the view vector
float3 R = (2 * cosNI) * bsdf->N - wi;
float3 R = (2 * cosNO) * bsdf->N - I;
float3 T, B;
make_orthonormals(R, &T, &B);
float phi = M_2PI_F * randu;
float cosTheta = powf(randv, 1 / (m_exponent + 1));
float sinTheta2 = 1 - cosTheta * cosTheta;
float sinTheta = sinTheta2 > 0 ? sqrtf(sinTheta2) : 0;
*wo = (cosf(phi) * sinTheta) * T + (sinf(phi) * sinTheta) * B + (cosTheta)*R;
if (dot(Ng, *wo) > 0.0f) {
*omega_in = (cosf(phi) * sinTheta) * T + (sinf(phi) * sinTheta) * B + (cosTheta)*R;
if (dot(Ng, *omega_in) > 0.0f) {
// common terms for pdf and eval
float cosNO = dot(bsdf->N, *wo);
float cosNI = dot(bsdf->N, *omega_in);
// make sure the direction we chose is still in the right hemisphere
if (cosNO > 0) {
if (cosNI > 0) {
float cosp = powf(cosTheta, m_exponent);
float common = 0.5f * M_1_PI_F * cosp;
*pdf = (m_exponent + 1) * common;
float out = cosNO * (m_exponent + 2) * common;
float out = cosNI * (m_exponent + 2) * common;
*eval = rgb_to_spectrum(bsdf_phong_ramp_get_color(bsdf->colors, cosp) * out);
}
}

Some files were not shown because too many files have changed in this diff Show More