First commit to this branch

This branch is supposed to serve as a template and tutorial to
newcomers of the Blender codebase.

CMakeLists.txt

@@ -1446,7 +1446,6 @@ elseif(CMAKE_C_COMPILER_ID MATCHES "MSVC")
 		# warning level:
 		"/W3"
 		"/w34062"  # switch statement contains 'default' but no 'case' labels
-		"/w34115"  # 'type' : named type definition in parentheses
 		"/w34189"  # local variable is initialized but not referenced
 		# disable:
 		"/wd4018"  # signed/unsigned mismatch

build_files/build_environment/CMakeLists.txt

@@ -25,16 +25,16 @@
 # Windows and macOS, and install_deps.sh on Linux.
 #
 # WINDOWS USAGE:
-#   Don't call this cmake file yourself, use build_deps.cmd
+#   Don't call this cmake file your self, use build_deps.cmd
 #   build_deps 2013 x64 / build_deps 2013 x86
 #   build_deps 2015 x64 / build_deps 2015 x86
 #
 # MAC OS X USAGE:
-#   Install with homebrew: brew install cmake autoconf automake libtool yasm nasm
+#	Install with homebrew: brew install autoconf automake libtool yasm nasm
 #   Run "make deps" from main Blender directory
 #
 # LINUX USAGE:
-#   Install compiler cmake autoconf automake libtool yasm nasm tcl
+#   Install compiler, cmake, autoconf, automake, libtool, yasm
 #   Run "make deps" from main Blender directory
 #
 ####################################################################################################

build_files/build_environment/cmake/harvest.cmake

@@ -129,8 +129,6 @@ harvest(jpg/include jpeg/include "*.h")
 harvest(jpg/lib jpeg/lib "libjpeg.a")
 harvest(lame/lib ffmpeg/lib "*.a")
 harvest(clang/bin llvm/bin "clang-format")
-harvest(clang/include llvm/include "*")
-harvest(llvm/include llvm/include "*")
 harvest(llvm/bin llvm/bin "llvm-config")
 harvest(llvm/lib llvm/lib "libLLVM*.a")
 if(APPLE)

build_files/build_environment/cmake/llvm.cmake

@@ -47,8 +47,7 @@ if(MSVC)
 		set(LLVM_HARVEST_COMMAND ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/ ${HARVEST_TARGET}/llvm/ )
 	else()
 		set(LLVM_HARVEST_COMMAND
-			${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/lib/ ${HARVEST_TARGET}/llvm/debug/lib/ &&
-			${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/include/ ${HARVEST_TARGET}/llvm/debug/include/
+			${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/lib/ ${HARVEST_TARGET}/llvm/debug/lib/
 		)
 	endif()
 	ExternalProject_Add_Step(ll after_install

build_files/build_environment/cmake/openal.cmake

@@ -18,31 +18,18 @@
 
 if(BUILD_MODE STREQUAL Release)
 	set(OPENAL_EXTRA_ARGS
-		-DALSOFT_UTILS=OFF
-		-DALSOFT_NO_CONFIG_UTIL=ON
-		-DALSOFT_EXAMPLES=OFF
-		-DALSOFT_TESTS=OFF
-		-DALSOFT_CONFIG=OFF
-		-DALSOFT_HRTF_DEFS=OFF
-		-DALSOFT_INSTALL=ON
-		-DALSOFT_BACKEND_SNDIO=OFF
+		-DALSOFT_UTILS=Off
+		-DALSOFT_NO_CONFIG_UTIL=On
+		-DALSOFT_EXAMPLES=Off
+		-DALSOFT_TESTS=Off
+		-DALSOFT_CONFIG=Off
+		-DALSOFT_HRTF_DEFS=Off
+		-DALSOFT_INSTALL=On
+		-DALSOFT_BACKEND_SNDIO=Off
 	)
 
 	if(UNIX)
-		set(OPENAL_EXTRA_ARGS
-			${OPENAL_EXTRA_ARGS}
-			-DLIBTYPE=STATIC
-		)
-	endif()
-
-	if(UNIX AND NOT APPLE)
-		# Ensure we have backends for playback.
-		set(OPENAL_EXTRA_ARGS
-			${OPENAL_EXTRA_ARGS}
-			-DALSOFT_REQUIRE_ALSA=ON
-			-DALSOFT_REQUIRE_OSS=ON
-			-DALSOFT_REQUIRE_PULSEAUDIO=ON
-		)
+		set(OPENAL_EXTRA_ARGS ${OPENAL_EXTRA_ARGS} -DLIBTYPE=STATIC)
 	endif()
 
 	ExternalProject_Add(external_openal

build_files/build_environment/cmake/versions.cmake

@@ -71,7 +71,7 @@ else()
 endif()
 
 set(FREETYPE_VERSION 2.9.1)
-set(FREETYPE_URI http://prdownloads.sourceforge.net/freetype/freetype-${FREETYPE_VERSION}.tar.gz)
+set(FREETYPE_URI http://download.savannah.gnu.org/releases/freetype/freetype-${FREETYPE_VERSION}.tar.gz)
 set(FREETYPE_HASH 3adb0e35d3c100c456357345ccfa8056)
 
 set(GLEW_VERSION 1.13.0)
@@ -251,7 +251,7 @@ set(JEMALLOC_URI https://github.com/jemalloc/jemalloc/releases/download/${JEMALL
 set(JEMALLOC_HASH 507f7b6b882d868730d644510491d18f)
 
 set(XML2_VERSION 2.9.4)
-set(XML2_URI http://xmlsoft.org/sources/libxml2-${XML2_VERSION}.tar.gz)
+set(XML2_URI ftp://xmlsoft.org/libxml2/libxml2-${XML2_VERSION}.tar.gz)
 set(XML2_HASH ae249165c173b1ff386ee8ad676815f5)
 
 set(TINYXML_VERSION 2_6_2)
@@ -284,7 +284,7 @@ set(BZIP2_URI http://http.debian.net/debian/pool/main/b/bzip2/bzip2_${BZIP2_VERS
 set(BZIP2_HASH d70a9ccd8bdf47e302d96c69fecd54925f45d9c7b966bb4ef5f56b770960afa7)
 
 set(FFI_VERSION 3.2.1)
-set(FFI_URI https://sourceware.org/pub/libffi/libffi-${FFI_VERSION}.tar.gz)
+set(FFI_URI ftp://sourceware.org/pub/libffi/libffi-${FFI_VERSION}.tar.gz)
 set(FFI_HASH d06ebb8e1d9a22d19e38d63fdb83954253f39bedc5d46232a05645685722ca37)
 
 set(LZMA_VERSION 5.2.4)

build_files/build_environment/install_deps.sh

@@ -301,8 +301,6 @@ NO_BUILD=false
 NO_CONFIRM=false
 USE_CXX11=true
 
-CLANG_FORMAT_VERSION_MIN="6.0"
-
 PYTHON_VERSION="3.7.0"
 PYTHON_VERSION_MIN="3.7"
 PYTHON_FORCE_BUILD=false
@@ -385,6 +383,7 @@ OPENCOLLADA_FORCE_BUILD=false
 OPENCOLLADA_FORCE_REBUILD=false
 OPENCOLLADA_SKIP=false
 
+
 EMBREE_VERSION="3.2.4"
 EMBREE_FORCE_BUILD=false
 EMBREE_FORCE_REBUILD=false
@@ -2629,7 +2628,7 @@ compile_FFmpeg() {
     ./configure --cc="gcc -Wl,--as-needed" \
         --extra-ldflags="-pthread -static-libgcc" \
         --prefix=$_inst --enable-static \
-        --disable-ffplay --disable-doc \
+        --disable-ffplay --disable-ffserver --disable-doc \
         --enable-gray \
         --enable-avfilter --disable-vdpau \
         --disable-bzlib --disable-libgsm --disable-libspeex \
@@ -2789,17 +2788,6 @@ install_DEB() {
     PRINT ""
   fi
 
-  PRINT ""
-  CLANG_FORMAT="clang-format"
-  check_package_version_ge_DEB $CLANG_FORMAT $CLANG_FORMAT_VERSION_MIN
-  if [ $? -eq 0 ]; then
-    _packages="$_packages $CLANG_FORMAT"
-  else
-    PRINT ""
-    WARNING "clang-format $CLANG_FORMAT_VERSION_MIN or higher not found, this is NOT needed to get Blender compiling..."
-    PRINT ""
-  fi
-
   if [ "$WITH_JACK" = true ]; then
     _packages="$_packages libspnav-dev"
     # Only install jack if jack2 is not already installed!
@@ -3451,16 +3439,6 @@ install_RPM() {
     install_packages_RPM libspnav-devel
   fi
 
-  PRINT ""
-  CLANG_FORMAT="clang"  # Yeah, on fedora/suse clang-format is part of main clang package...
-  check_package_version_ge_RPM $CLANG_FORMAT $CLANG_FORMAT_VERSION_MIN
-  if [ $? -eq 0 ]; then
-    install_packages_RPM $CLANG_FORMAT
-  else
-    PRINT ""
-    WARNING "clang-format $CLANG_FORMAT_VERSION_MIN or higher not found, this is NOT needed to get Blender compiling..."
-    PRINT ""
-  fi
 
   PRINT ""
   _do_compile_python=false
@@ -3908,18 +3886,6 @@ install_ARCH() {
   fi
 
 
-  PRINT ""
-  CLANG_FORMAT="clang"  # Yeah, on arch clang-format is part of main clang package...
-  check_package_version_ge_ARCH $CLANG_FORMAT $CLANG_FORMAT_VERSION_MIN
-  if [ $? -eq 0 ]; then
-    install_packages_ARCH $CLANG_FORMAT
-  else
-    PRINT ""
-    WARNING "clang-format $CLANG_FORMAT_VERSION_MIN or higher not found, this is NOT needed to get Blender compiling..."
-    PRINT ""
-  fi
-
-
   PRINT ""
   _do_compile_python=false
   if [ "$PYTHON_SKIP" = true ]; then
@@ -4558,33 +4524,18 @@ print_info() {
       PRINT "  $_1"
       _buildargs="$_buildargs $_1"
     fi
-    if [ -d $INST/blosc ]; then
-      _1="-D BLOSC_ROOT_DIR=$INST/blosc"
-      PRINT "  $_1"
-      _buildargs="$_buildargs $_1"
-    fi
   fi
 
   if [ "$WITH_OPENCOLLADA" = true ]; then
     _1="-D WITH_OPENCOLLADA=ON"
     PRINT "  $_1"
     _buildargs="$_buildargs $_1"
-    if [ -d $INST/opencollada ]; then
-      _1="-D OPENCOLLADA_ROOT_DIR=$INST/opencollada"
-      PRINT "  $_1"
-      _buildargs="$_buildargs $_1"
-    fi
   fi
 
   if [ "$WITH_EMBREE" = true ]; then
     _1="-D WITH_CYCLES_EMBREE=ON"
     PRINT "  $_1"
     _buildargs="$_buildargs $_1"
-    if [ -d $INST/embree ]; then
-      _1="-D EMBREE_ROOT_DIR=$INST/embree"
-      PRINT "  $_1"
-      _buildargs="$_buildargs $_1"
-    fi
   fi
 
   if [ "$WITH_JACK" = true ]; then

build_files/buildbot/slave_compile.py

@@ -54,15 +54,18 @@ if 'cmake' in builder:
     targets = ['blender']
 
     chroot_name = None  # If not None command will be delegated to that chroot
+    cuda_chroot_name = None  # If not None cuda compilationcommand will be delegated to that chroot
     build_cubins = True  # Whether to build Cycles CUDA kernels
     bits = 64
 
     # Config file to be used (relative to blender's sources root)
     cmake_config_file = "build_files/cmake/config/blender_release.cmake"
+    cmake_cuda_config_file = None
 
     # Set build options.
     cmake_options = []
     cmake_extra_options = ['-DCMAKE_BUILD_TYPE:STRING=Release']
+    cuda_cmake_options = []
 
     if builder.startswith('mac'):
         # Set up OSX architecture
@@ -71,11 +74,26 @@ if 'cmake' in builder:
         cmake_extra_options.append('-DCMAKE_OSX_DEPLOYMENT_TARGET=10.9')
 
     elif builder.startswith('win'):
-        if builder.startswith('win64'):
-            cmake_options.extend(['-G', 'Visual Studio 15 2017 Win64'])
-        elif builder.startswith('win32'):
-            bits = 32
-            cmake_options.extend(['-G', 'Visual Studio 15 2017'])
+        if builder.endswith('_vs2017'):
+            if builder.startswith('win64'):
+                cmake_options.extend(['-G', 'Visual Studio 15 2017 Win64'])
+            elif builder.startswith('win32'):
+                bits = 32
+                cmake_options.extend(['-G', 'Visual Studio 15 2017'])
+        elif builder.endswith('_vc2015'):
+            if builder.startswith('win64'):
+                cmake_options.extend(['-G', 'Visual Studio 14 2015 Win64'])
+            elif builder.startswith('win32'):
+                bits = 32
+                cmake_options.extend(['-G', 'Visual Studio 14 2015'])
+            cmake_extra_options.append('-DCUDA_NVCC_FLAGS=--cl-version;2013;' +
+                '--compiler-bindir;C:\\Program Files (x86)\\Microsoft Visual Studio 12.0\\VC\\bin')
+        else:
+            if builder.startswith('win64'):
+                cmake_options.extend(['-G', 'Visual Studio 12 2013 Win64'])
+            elif builder.startswith('win32'):
+                bits = 32
+                cmake_options.extend(['-G', 'Visual Studio 12 2013'])
 
     elif builder.startswith('linux'):
         tokens = builder.split("_")
@@ -98,14 +116,36 @@ if 'cmake' in builder:
             cmake_extra_options.extend(["-DCMAKE_C_COMPILER=/usr/bin/gcc-7",
                                         "-DCMAKE_CXX_COMPILER=/usr/bin/g++-7"])
 
+    # Workaround to build only sm_7x kernels with CUDA 10, until
+    # older kernels work well with this version.
+    if builder.startswith('win'):
+        cmake_extra_options.append('-DCUDA_VERSION=9.1')
+        cmake_extra_options.append('-DCUDA_TOOLKIT_INCLUDE:PATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v9.1/include')
+        cmake_extra_options.append('-DCUDA_TOOLKIT_ROOT_DIR:PATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v9.1')
+        cmake_extra_options.append('-DCUDA_NVCC_EXECUTABLE:FILEPATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v9.1/bin/nvcc.exe')
+        cmake_extra_options.append('-DCUDA10_NVCC_EXECUTABLE:FILEPATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0/bin/nvcc.exe')
+        cmake_extra_options.append('-DCUDA10_TOOLKIT_ROOT_DIR:PATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0')
+    elif builder.startswith('linux'):
+        cmake_extra_options.append('-DCUDA_VERSION=9.1')
+        cmake_extra_options.append('-DCUDA_TOOLKIT_INCLDUE:PATH=/usr/local/cuda-9.1/include')
+        cmake_extra_options.append('-DCUDA_TOOLKIT_ROOT_DIR:PATH=/usr/local/cuda-9.1')
+        cmake_extra_options.append('-DCUDA_NVCC_EXECUTABLE:FILEPATH=/usr/local/cuda-9.1/bin/nvcc')
+        cmake_extra_options.append('-DCUDA10_NVCC_EXECUTABLE:FILEPATH=/usr/local/cuda-10.0/bin/nvcc')
+        cmake_extra_options.append('-DCUDA10_TOOLKIT_ROOT_DIR:PATH=/usr/local/cuda-10.0')
+
     cmake_options.append("-C" + os.path.join(blender_dir, cmake_config_file))
 
     # Prepare CMake options needed to configure cuda binaries compilation, 64bit only.
-    if bits == 64 and build_cubins:
-        cmake_options.append("-DWITH_CYCLES_CUDA_BINARIES=ON")
-        cmake_options.append("-DCUDA_64_BIT_DEVICE_CODE=ON")
+    if bits == 64:
+        cuda_cmake_options.append("-DWITH_CYCLES_CUDA_BINARIES=%s" % ('ON' if build_cubins else 'OFF'))
+        if build_cubins or 'cuda' in targets:
+            cuda_cmake_options.append("-DCUDA_64_BIT_DEVICE_CODE=ON")
+
+        # Only modify common cmake options if cuda doesn't require separate target.
+        if 'cuda' not in targets:
+            cmake_options += cuda_cmake_options
     else:
-        cmake_options.append("-DWITH_CYCLES_CUDA_BINARIES=OFF")
+        cuda_cmake_options.append("-DWITH_CYCLES_CUDA_BINARIES=OFF")
 
     cmake_options.append("-DCMAKE_INSTALL_PREFIX=%s" % (install_dir))
 
@@ -116,6 +156,10 @@ if 'cmake' in builder:
         chroot_prefix = ['schroot', '-c', chroot_name, '--']
     else:
         chroot_prefix = []
+    if cuda_chroot_name:
+        cuda_chroot_prefix = ['schroot', '-c', cuda_chroot_name, '--']
+    else:
+        cuda_chroot_prefix = chroot_prefix[:]
 
     # Make sure no garbage remained from the previous run
     if os.path.isdir(install_dir):
@@ -131,6 +175,14 @@ if 'cmake' in builder:
         target_name = 'install'
         # Tweaking CMake options to respect the target
         target_cmake_options = cmake_options[:]
+        if target == 'cuda':
+            target_cmake_options += cuda_cmake_options
+            target_chroot_prefix = cuda_chroot_prefix[:]
+            target_name = 'cycles_kernel_cuda'
+        # If cuda binaries are compiled as a separate target, make sure
+        # other targets don't compile cuda binaries.
+        if 'cuda' in targets and target != 'cuda':
+            target_cmake_options.append("-DWITH_CYCLES_CUDA_BINARIES=OFF")
         # Do extra git fetch because not all platform/git/buildbot combinations
         # update the origin remote, causing buildinfo to detect local changes.
         os.chdir(blender_dir)
@@ -167,6 +219,16 @@ if 'cmake' in builder:
         if retcode != 0:
             sys.exit(retcode)
 
+        if builder.startswith('linux') and target == 'cuda':
+            blender_h = os.path.join(blender_dir, "source", "blender", "blenkernel", "BKE_blender_version.h")
+            blender_version = int(parse_header_file(blender_h, 'BLENDER_VERSION'))
+            blender_version = "%d.%d" % (blender_version // 100, blender_version % 100)
+            kernels = os.path.join(target_build_dir, 'intern', 'cycles', 'kernel')
+            install_kernels = os.path.join(install_dir, blender_version, 'scripts', 'addons', 'cycles', 'lib')
+            os.mkdir(install_kernels)
+            print("Copying cuda binaries from %s to %s" % (kernels, install_kernels))
+            os.system('cp %s/*.cubin %s' % (kernels, install_kernels))
+
 else:
     print("Unknown building system")
     sys.exit(1)

build_files/cmake/config/blender_release.cmake

@@ -50,9 +50,8 @@ set(WITH_SDL                 ON  CACHE BOOL "" FORCE)
 set(WITH_X11_XINPUT          ON  CACHE BOOL "" FORCE)
 set(WITH_X11_XF86VMODE       ON  CACHE BOOL "" FORCE)
 
-set(WITH_MEM_JEMALLOC          ON  CACHE BOOL "" FORCE)
-set(WITH_CYCLES_CUDA_BINARIES  ON  CACHE BOOL "" FORCE)
-set(WITH_CYCLES_CUBIN_COMPILER OFF CACHE BOOL "" FORCE)
+set(WITH_MEM_JEMALLOC        ON  CACHE BOOL "" FORCE)
+set(WITH_CYCLES_CUDA_BINARIES ON  CACHE BOOL "" FORCE)
 set(CYCLES_CUDA_BINARIES_ARCH sm_30;sm_35;sm_37;sm_50;sm_52;sm_60;sm_61;sm_70;sm_75 CACHE STRING "" FORCE)
 
 # platform dependent options

build_files/cmake/config/bpy_module.cmake

@@ -30,8 +30,3 @@ set(WITH_INTERNATIONAL       OFF CACHE BOOL "" FORCE)
 set(WITH_BULLET              OFF CACHE BOOL "" FORCE)
 set(WITH_OPENVDB             OFF CACHE BOOL "" FORCE)
 set(WITH_ALEMBIC             OFF CACHE BOOL "" FORCE)
-
-if(CMAKE_SYSTEM_NAME MATCHES "Linux")
-	# jemalloc causes linking error on import, disable.
-	set(WITH_MEM_JEMALLOC        OFF CACHE BOOL "" FORCE)
-endif()

build_files/cmake/packaging.cmake

@@ -86,7 +86,7 @@ if(WIN32)
 	set(CPACK_NSIS_MUI_ICON ${CMAKE_SOURCE_DIR}/release/windows/icons/winblender.ico)
 	set(CPACK_NSIS_COMPRESSOR "/SOLID lzma")
 
-	set(CPACK_RESOURCE_FILE_LICENSE ${CMAKE_SOURCE_DIR}/release/text/GPL3-license.txt)
+	set(CPACK_RESOURCE_FILE_LICENSE ${CMAKE_SOURCE_DIR}/release/text/GPL-license.txt)
 	set(CPACK_WIX_PRODUCT_ICON ${CMAKE_SOURCE_DIR}/release/windows/icons/winblender.ico)
 	set(CPACK_WIX_UPGRADE_GUID "B767E4FD-7DE7-4094-B051-3AE62E13A17A")
 

build_files/cmake/platform/platform_win32.cmake

@@ -427,7 +427,6 @@ 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")
 	file(GLOB LLVM_LIBRARY_OPTIMIZED ${LLVM_ROOT_DIR}/lib/*.lib)
 
 	if(EXISTS ${LLVM_ROOT_DIR}/debug/lib)

doc/blender_file_format/BlendFileDnaExporter_25.py

@@ -57,6 +57,7 @@
 
 import struct
 import sys
+import getopt                   # command line arguments handling
 from string import Template     # strings completion
 
 

doc/doxygen/doxygen.intern.h

@@ -71,7 +71,7 @@
  *  \ingroup audaspace
  */
 /** \defgroup audsrc Audaspace SRC
- *
+
  *  \ingroup audaspace
  */
 /** \defgroup audffmpeg Audaspace FFMpeg
@@ -91,3 +91,4 @@
  * \ingroup intern GUI
  * \ref GHOSTPage
  */
+

doc/python_api/examples/blf.py

@@ -6,6 +6,7 @@ Example of using the blf module. For this module to work we
 need to use the OpenGL wrapper :class:`~bgl` as well.
 """
 # import stand alone modules
+import bgl
 import blf
 import bpy
 

doc/python_api/examples/bpy.props.1.py

@@ -15,15 +15,13 @@ class OBJECT_OT_property_example(bpy.types.Operator):
     bl_label = "Property Example"
     bl_options = {'REGISTER', 'UNDO'}
 
-    my_float: bpy.props.FloatProperty(name="Some Floating Point")
-    my_bool: bpy.props.BoolProperty(name="Toggle Option")
-    my_string: bpy.props.StringProperty(name="String Value")
+    my_float = bpy.props.FloatProperty(name="Some Floating Point")
+    my_bool = bpy.props.BoolProperty(name="Toggle Option")
+    my_string = bpy.props.StringProperty(name="String Value")
 
     def execute(self, context):
-        self.report(
-            {'INFO'}, 'F: %.2f  B: %s  S: %r' %
-            (self.my_float, self.my_bool, self.my_string)
-        )
+        self.report({'INFO'}, 'F: %.2f  B: %s  S: %r' %
+                    (self.my_float, self.my_bool, self.my_string))
         print('My float:', self.my_float)
         print('My bool:', self.my_bool)
         print('My string:', self.my_string)
@@ -55,8 +53,5 @@ bpy.utils.register_class(OBJECT_OT_property_example)
 bpy.utils.register_class(OBJECT_PT_property_example)
 
 # Demo call. Be sure to also test in the 3D Viewport.
-bpy.ops.object.property_example(
-    my_float=47,
-    my_bool=True,
-    my_string="Shouldn't that be 327?",
-)
+bpy.ops.object.property_example(my_float=47, my_bool=True,
+                                my_string="Shouldn't that be 327?")

doc/python_api/examples/bpy.props.2.py

@@ -10,14 +10,15 @@ import bpy
 
 
 class MaterialSettings(bpy.types.PropertyGroup):
-    my_int: bpy.props.IntProperty()
-    my_float: bpy.props.FloatProperty()
-    my_string: bpy.props.StringProperty()
+    my_int = bpy.props.IntProperty()
+    my_float = bpy.props.FloatProperty()
+    my_string = bpy.props.StringProperty()
 
 
 bpy.utils.register_class(MaterialSettings)
 
-bpy.types.Material.my_settings = bpy.props.PointerProperty(type=MaterialSettings)
+bpy.types.Material.my_settings = \
+    bpy.props.PointerProperty(type=MaterialSettings)
 
 # test the new settings work
 material = bpy.data.materials[0]

doc/python_api/examples/bpy.props.3.py

@@ -9,17 +9,18 @@ Custom properties can be added to any subclass of an :class:`ID`,
 import bpy
 
 
-# Assign a collection.
+# Assign a collection
 class SceneSettingItem(bpy.types.PropertyGroup):
-    name: bpy.props.StringProperty(name="Test Property", default="Unknown")
-    value: bpy.props.IntProperty(name="Test Property", default=22)
+    name = bpy.props.StringProperty(name="Test Prop", default="Unknown")
+    value = bpy.props.IntProperty(name="Test Prop", default=22)
 
 
 bpy.utils.register_class(SceneSettingItem)
 
-bpy.types.Scene.my_settings = bpy.props.CollectionProperty(type=SceneSettingItem)
+bpy.types.Scene.my_settings = \
+    bpy.props.CollectionProperty(type=SceneSettingItem)
 
-# Assume an armature object selected.
+# Assume an armature object selected
 print("Adding 2 values!")
 
 my_item = bpy.context.scene.my_settings.add()

doc/python_api/examples/bpy.props.py

@@ -12,7 +12,7 @@ like Blender's existing properties.
 import bpy
 
 # Assign a custom property to an existing type.
-bpy.types.Material.custom_float = bpy.props.FloatProperty(name="Test Property")
+bpy.types.Material.custom_float = bpy.props.FloatProperty(name="Test Prob")
 
 # Test the property is there.
 bpy.data.materials[0].custom_float = 5.0

doc/python_api/examples/bpy.types.AddonPreferences.1.py

@@ -22,15 +22,15 @@ class ExampleAddonPreferences(AddonPreferences):
     # when defining this in a submodule of a python package.
     bl_idname = __name__
 
-    filepath: StringProperty(
+    filepath = StringProperty(
         name="Example File Path",
         subtype='FILE_PATH',
     )
-    number: IntProperty(
+    number = IntProperty(
         name="Example Number",
         default=4,
     )
-    boolean: BoolProperty(
+    boolean = BoolProperty(
         name="Example Boolean",
         default=False,
     )

doc/python_api/examples/bpy.types.Object.py

@@ -7,6 +7,7 @@ object, placing it into a view layer, selecting it and making it active.
 """
 
 import bpy
+from mathutils import Matrix
 
 view_layer = bpy.context.view_layer
 

doc/python_api/examples/bpy.types.Operator.1.py

@@ -28,8 +28,8 @@ class SimpleMouseOperator(bpy.types.Operator):
     bl_idname = "wm.mouse_position"
     bl_label = "Invoke Mouse Operator"
 
-    x: bpy.props.IntProperty()
-    y: bpy.props.IntProperty()
+    x = bpy.props.IntProperty()
+    y = bpy.props.IntProperty()
 
     def execute(self, context):
         # rather than printing, use the report function,

doc/python_api/examples/bpy.types.Operator.2.py

@@ -21,7 +21,7 @@ class ExportSomeData(bpy.types.Operator):
     bl_idname = "export.some_data"
     bl_label = "Export Some Data"
 
-    filepath: bpy.props.StringProperty(subtype="FILE_PATH")
+    filepath = bpy.props.StringProperty(subtype="FILE_PATH")
 
     @classmethod
     def poll(cls, context):

doc/python_api/examples/bpy.types.Operator.3.py

@@ -11,15 +11,13 @@ class DialogOperator(bpy.types.Operator):
     bl_idname = "object.dialog_operator"
     bl_label = "Simple Dialog Operator"
 
-    my_float: bpy.props.FloatProperty(name="Some Floating Point")
-    my_bool: bpy.props.BoolProperty(name="Toggle Option")
-    my_string: bpy.props.StringProperty(name="String Value")
+    my_float = bpy.props.FloatProperty(name="Some Floating Point")
+    my_bool = bpy.props.BoolProperty(name="Toggle Option")
+    my_string = bpy.props.StringProperty(name="String Value")
 
     def execute(self, context):
-        message = (
-            "Popup Values: %f, %d, '%s'" %
+        message = "Popup Values: %f, %d, '%s'" % \
             (self.my_float, self.my_bool, self.my_string)
-        )
         self.report({'INFO'}, message)
         return {'FINISHED'}
 
@@ -30,5 +28,5 @@ class DialogOperator(bpy.types.Operator):
 
 bpy.utils.register_class(DialogOperator)
 
-# Test call.
+# test call
 bpy.ops.object.dialog_operator('INVOKE_DEFAULT')

doc/python_api/examples/bpy.types.Operator.4.py

@@ -16,11 +16,11 @@ class CustomDrawOperator(bpy.types.Operator):
     bl_idname = "object.custom_draw"
     bl_label = "Simple Modal Operator"
 
-    filepath: bpy.props.StringProperty(subtype="FILE_PATH")
+    filepath = bpy.props.StringProperty(subtype="FILE_PATH")
 
-    my_float: bpy.props.FloatProperty(name="Float")
-    my_bool: bpy.props.BoolProperty(name="Toggle Option")
-    my_string: bpy.props.StringProperty(name="String Value")
+    my_float = bpy.props.FloatProperty(name="Float")
+    my_bool = bpy.props.BoolProperty(name="Toggle Option")
+    my_string = bpy.props.StringProperty(name="String Value")
 
     def execute(self, context):
         print("Test", self)

doc/python_api/examples/bpy.types.Operator.6.py

@@ -14,7 +14,7 @@ class SearchEnumOperator(bpy.types.Operator):
     bl_label = "Search Enum Operator"
     bl_property = "my_search"
 
-    my_search: EnumProperty(
+    my_search = EnumProperty(
         name="My Search",
         items=(
             ('FOO', "Foo", ""),

doc/python_api/examples/bpy.types.PropertyGroup.py

@@ -28,8 +28,8 @@ import bpy
 
 
 class MyPropertyGroup(bpy.types.PropertyGroup):
-    custom_1: bpy.props.FloatProperty(name="My Float")
-    custom_2: bpy.props.IntProperty(name="My Int")
+    custom_1 = bpy.props.FloatProperty(name="My Float")
+    custom_2 = bpy.props.IntProperty(name="My Int")
 
 
 bpy.utils.register_class(MyPropertyGroup)

doc/python_api/examples/bpy.types.RenderEngine.py

@@ -4,218 +4,82 @@ Simple Render Engine
 """
 
 import bpy
-import bgl
 
 
 class CustomRenderEngine(bpy.types.RenderEngine):
     # These three members are used by blender to set up the
     # RenderEngine; define its internal name, visible name and capabilities.
-    bl_idname = "CUSTOM"
-    bl_label = "Custom"
+    bl_idname = "custom_renderer"
+    bl_label = "Flat Color Renderer"
     bl_use_preview = True
 
-    # Init is called whenever a new render engine instance is created. Multiple
-    # instances may exist at the same time, for example for a viewport and final
-    # render.
-    def __init__(self):
-        self.scene_data = None
-        self.draw_data = None
-
-    # When the render engine instance is destroy, this is called. Clean up any
-    # render engine data here, for example stopping running render threads.
-    def __del__(self):
-        pass
-
-    # This is the method called by Blender for both final renders (F12) and
-    # small preview for materials, world and lights.
-    def render(self, depsgraph):
-        scene = depsgraph.scene
+    # This is the only method called by blender, in this example
+    # we use it to detect preview rendering and call the implementation
+    # in another method.
+    def render(self, scene):
         scale = scene.render.resolution_percentage / 100.0
         self.size_x = int(scene.render.resolution_x * scale)
         self.size_y = int(scene.render.resolution_y * scale)
 
-        # Fill the render result with a flat color. The framebuffer is
-        # defined as a list of pixels, each pixel itself being a list of
-        # R,G,B,A values.
         if self.is_preview:
-            color = [0.1, 0.2, 0.1, 1.0]
+            self.render_preview(scene)
         else:
-            color = [0.2, 0.1, 0.1, 1.0]
+            self.render_scene(scene)
 
+    # In this example, we fill the preview renders with a flat green color.
+    def render_preview(self, scene):
         pixel_count = self.size_x * self.size_y
-        rect = [color] * pixel_count
+
+        # The framebuffer is defined as a list of pixels, each pixel
+        # itself being a list of R,G,B,A values
+        green_rect = [[0.0, 1.0, 0.0, 1.0]] * pixel_count
 
         # Here we write the pixel values to the RenderResult
         result = self.begin_result(0, 0, self.size_x, self.size_y)
         layer = result.layers[0].passes["Combined"]
-        layer.rect = rect
+        layer.rect = green_rect
         self.end_result(result)
 
-    # For viewport renders, this method gets called once at the start and
-    # whenever the scene or 3D viewport changes. This method is where data
-    # should be read from Blender in the same thread. Typically a render
-    # thread will be started to do the work while keeping Blender responsive.
-    def view_update(self, context):
-        region = context.region
-        view3d = context.space_data
-        depsgraph = context.depsgraph
-        scene = depsgraph.scene
+    # In this example, we fill the full renders with a flat blue color.
+    def render_scene(self, scene):
+        pixel_count = self.size_x * self.size_y
 
-        # Get viewport dimensions
-        dimensions = region.width, region.height
+        # The framebuffer is defined as a list of pixels, each pixel
+        # itself being a list of R,G,B,A values
+        blue_rect = [[0.0, 0.0, 1.0, 1.0]] * pixel_count
 
-        if not self.scene_data:
-            # First time initialization
-            self.scene_data = []
-            first_time = True
+        # Here we write the pixel values to the RenderResult
+        result = self.begin_result(0, 0, self.size_x, self.size_y)
+        layer = result.layers[0].passes["Combined"]
+        layer.rect = blue_rect
+        self.end_result(result)
 
-            # Loop over all datablocks used in the scene.
-            for datablock in depsgraph.ids:
-                pass
-        else:
-            first_time = False
-
-            # Test which datablocks changed
-            for update in depsgraph.updates:
-                print("Datablock updated: ", update.id.name)
-
-            # Test if any material was added, removed or changed.
-            if depsgraph.id_type_update('MATERIAL'):
-                print("Materials updated")
-
-        # Loop over all object instances in the scene.
-        if first_time or depsgraph.id_type_update('OBJECT'):
-            for instance in depsgraph.object_instances:
-                pass
-
-    # For viewport renders, this method is called whenever Blender redraws
-    # the 3D viewport. The renderer is expected to quickly draw the render
-    # with OpenGL, and not perform other expensive work.
-    # Blender will draw overlays for selection and editing on top of the
-    # rendered image automatically.
-    def view_draw(self, context):
-        region = context.region
-        depsgraph = context.depsgraph
-        scene = depsgraph.scene
-
-        # Get viewport dimensions
-        dimensions = region.width, region.height
-
-        # Bind shader that converts from scene linear to display space,
-        bgl.glEnable(bgl.GL_BLEND)
-        bgl.glBlendFunc(bgl.GL_ONE, bgl.GL_ONE_MINUS_SRC_ALPHA);
-        self.bind_display_space_shader(scene)
-
-        if not self.draw_data or self.draw_data.dimensions != dimensions:
-            self.draw_data = CustomDrawData(dimensions)
-
-        self.draw_data.draw()
-
-        self.unbind_display_space_shader()
-        bgl.glDisable(bgl.GL_BLEND)
-
-
-class CustomDrawData:
-    def __init__(self, dimensions):
-        # Generate dummy float image buffer
-        self.dimensions = dimensions
-        width, height = dimensions
-
-        pixels = [0.1, 0.2, 0.1, 1.0] * width * height
-        pixels = bgl.Buffer(bgl.GL_FLOAT, width * height * 4, pixels)
-
-        # Generate texture
-        self.texture = bgl.Buffer(bgl.GL_INT, 1)
-        bgl.glGenTextures(1, self.texture)
-        bgl.glActiveTexture(bgl.GL_TEXTURE0)
-        bgl.glBindTexture(bgl.GL_TEXTURE_2D, self.texture[0])
-        bgl.glTexImage2D(bgl.GL_TEXTURE_2D, 0, bgl.GL_RGBA16F, width, height, 0, bgl.GL_RGBA, bgl.GL_FLOAT, pixels)
-        bgl.glTexParameteri(bgl.GL_TEXTURE_2D, bgl.GL_TEXTURE_MIN_FILTER, bgl.GL_LINEAR)
-        bgl.glTexParameteri(bgl.GL_TEXTURE_2D, bgl.GL_TEXTURE_MAG_FILTER, bgl.GL_LINEAR)
-        bgl.glBindTexture(bgl.GL_TEXTURE_2D, 0)
-
-        # Bind shader that converts from scene linear to display space,
-        # use the scene's color management settings.
-        shader_program = bgl.Buffer(bgl.GL_INT, 1)
-        bgl.glGetIntegerv(bgl.GL_CURRENT_PROGRAM, shader_program);
-
-        # Generate vertex array
-        self.vertex_array = bgl.Buffer(bgl.GL_INT, 1)
-        bgl.glGenVertexArrays(1, self.vertex_array)
-        bgl.glBindVertexArray(self.vertex_array[0])
-
-        texturecoord_location = bgl.glGetAttribLocation(shader_program[0], "texCoord");
-        position_location = bgl.glGetAttribLocation(shader_program[0], "pos");
-
-        bgl.glEnableVertexAttribArray(texturecoord_location);
-        bgl.glEnableVertexAttribArray(position_location);
-
-        # Generate geometry buffers for drawing textured quad
-        position = [0.0, 0.0, width, 0.0, width, height, 0.0, height]
-        position = bgl.Buffer(bgl.GL_FLOAT, len(position), position)
-        texcoord = [0.0, 0.0, 1.0, 0.0, 1.0, 1.0, 0.0, 1.0]
-        texcoord = bgl.Buffer(bgl.GL_FLOAT, len(texcoord), texcoord)
-
-        self.vertex_buffer = bgl.Buffer(bgl.GL_INT, 2)
-
-        bgl.glGenBuffers(2, self.vertex_buffer)
-        bgl.glBindBuffer(bgl.GL_ARRAY_BUFFER, self.vertex_buffer[0])
-        bgl.glBufferData(bgl.GL_ARRAY_BUFFER, 32, position, bgl.GL_STATIC_DRAW)
-        bgl.glVertexAttribPointer(position_location, 2, bgl.GL_FLOAT, bgl.GL_FALSE, 0, None)
-
-        bgl.glBindBuffer(bgl.GL_ARRAY_BUFFER, self.vertex_buffer[1])
-        bgl.glBufferData(bgl.GL_ARRAY_BUFFER, 32, texcoord, bgl.GL_STATIC_DRAW)
-        bgl.glVertexAttribPointer(texturecoord_location, 2, bgl.GL_FLOAT, bgl.GL_FALSE, 0, None)
-
-        bgl.glBindBuffer(bgl.GL_ARRAY_BUFFER, 0)
-        bgl.glBindVertexArray(0)
-
-    def __del__(self):
-        bgl.glDeleteBuffers(2, self.vertex_buffer)
-        bgl.glDeleteVertexArrays(1, self.vertex_array)
-        bgl.glBindTexture(bgl.GL_TEXTURE_2D, 0)
-        bgl.glDeleteTextures(1, self.texture)
-
-    def draw(self):
-        bgl.glActiveTexture(bgl.GL_TEXTURE0)
-        bgl.glBindTexture(bgl.GL_TEXTURE_2D, self.texture[0])
-        bgl.glBindVertexArray(self.vertex_array[0])
-        bgl.glDrawArrays(bgl.GL_TRIANGLE_FAN, 0, 4);
-        bgl.glBindVertexArray(0)
-        bgl.glBindTexture(bgl.GL_TEXTURE_2D, 0)
-
-
-# RenderEngines also need to tell UI Panels that they are compatible with.
-# We recommend to enable all panels marked as BLENDER_RENDER, and then
-# exclude any panels that are replaced by custom panels registered by the
-# render engine, or that are not supported.
-def get_panels():
-    exclude_panels = {
-        'VIEWLAYER_PT_filter',
-        'VIEWLAYER_PT_layer_passes',
-    }
-
-    panels = []
-    for panel in bpy.types.Panel.__subclasses__():
-        if hasattr(panel, 'COMPAT_ENGINES') and 'BLENDER_RENDER' in panel.COMPAT_ENGINES:
-            if panel.__name__ not in exclude_panels:
-                panels.append(panel)
-
-    return panels
 
 def register():
     # Register the RenderEngine
     bpy.utils.register_class(CustomRenderEngine)
 
-    for panel in get_panels():
-        panel.COMPAT_ENGINES.add('CUSTOM')
+    # RenderEngines also need to tell UI Panels that they are compatible
+    # Otherwise most of the UI will be empty when the engine is selected.
+    # In this example, we need to see the main render image button and
+    # the material preview panel.
+    from bl_ui import (
+        properties_render,
+        properties_material,
+    )
+    properties_render.RENDER_PT_render.COMPAT_ENGINES.add(CustomRenderEngine.bl_idname)
+    properties_material.MATERIAL_PT_preview.COMPAT_ENGINES.add(CustomRenderEngine.bl_idname)
+
 
 def unregister():
     bpy.utils.unregister_class(CustomRenderEngine)
 
-    for panel in get_panels():
-        if 'CUSTOM' in panel.COMPAT_ENGINES:
-            panel.COMPAT_ENGINES.remove('CUSTOM')
+    from bl_ui import (
+        properties_render,
+        properties_material,
+    )
+    properties_render.RENDER_PT_render.COMPAT_ENGINES.remove(CustomRenderEngine.bl_idname)
+    properties_material.MATERIAL_PT_preview.COMPAT_ENGINES.remove(CustomRenderEngine.bl_idname)
 
 
 if __name__ == "__main__":

doc/python_api/examples/bpy.types.UIList.2.py

@@ -15,24 +15,12 @@ class MESH_UL_vgroups_slow(bpy.types.UIList):
     VGROUP_EMPTY = 1 << 0
 
     # Custom properties, saved with .blend file.
-    use_filter_empty: bpy.props.BoolProperty(
-        name="Filter Empty",
-        default=False,
-        options=set(),
-        description="Whether to filter empty vertex groups",
-    )
-    use_filter_empty_reverse: bpy.props.BoolProperty(
-        name="Reverse Empty",
-        default=False,
-        options=set(),
-        description="Reverse empty filtering",
-    )
-    use_filter_name_reverse: bpy.props.BoolProperty(
-        name="Reverse Name",
-        default=False,
-        options=set(),
-        description="Reverse name filtering",
-    )
+    use_filter_empty = bpy.props.BoolProperty(name="Filter Empty", default=False, options=set(),
+                                              description="Whether to filter empty vertex groups")
+    use_filter_empty_reverse = bpy.props.BoolProperty(name="Reverse Empty", default=False, options=set(),
+                                                      description="Reverse empty filtering")
+    use_filter_name_reverse = bpy.props.BoolProperty(name="Reverse Name", default=False, options=set(),
+                                                     description="Reverse name filtering")
 
     # This allows us to have mutually exclusive options, which are also all disable-able!
     def _gen_order_update(name1, name2):
@@ -40,18 +28,12 @@ class MESH_UL_vgroups_slow(bpy.types.UIList):
             if (getattr(self, name1)):
                 setattr(self, name2, False)
         return _u
-    use_order_name: bpy.props.BoolProperty(
-        name="Name", default=False, options=set(),
-        description="Sort groups by their name (case-insensitive)",
-        update=_gen_order_update("use_order_name", "use_order_importance"),
-    )
-    use_order_importance: bpy.props.BoolProperty(
-        name="Importance",
-        default=False,
-        options=set(),
-        description="Sort groups by their average weight in the mesh",
-        update=_gen_order_update("use_order_importance", "use_order_name"),
-    )
+    use_order_name = bpy.props.BoolProperty(name="Name", default=False, options=set(),
+                                            description="Sort groups by their name (case-insensitive)",
+                                            update=_gen_order_update("use_order_name", "use_order_importance"))
+    use_order_importance = bpy.props.BoolProperty(name="Importance", default=False, options=set(),
+                                                  description="Sort groups by their average weight in the mesh",
+                                                  update=_gen_order_update("use_order_importance", "use_order_name"))
 
     # Usual draw item function.
     def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index, flt_flag):

doc/python_api/examples/bpy.types.bpy_struct.keyframe_insert.1.py

@@ -5,32 +5,25 @@ than the bone.
 """
 
 import bpy
-from bpy.props import (
-    FloatProperty,
-    PointerProperty,
-)
+from bpy.props import PointerProperty
 
 
-# Define a nested property.
+# define a nested property
 class MyPropGroup(bpy.types.PropertyGroup):
-    nested: FloatProperty(name="Nested", default=0.0)
+    nested = bpy.props.FloatProperty(name="Nested", default=0.0)
 
 
-# Register it so its available for all bones.
+# register it so its available for all bones
 bpy.utils.register_class(MyPropGroup)
-bpy.types.Bone.my_prop = PointerProperty(
-    type=MyPropGroup,
-    name="MyProp",
-)
+bpy.types.Bone.my_prop = PointerProperty(type=MyPropGroup,
+                                         name="MyProp")
 
-# Get a bone.
+# get a bone
 obj = bpy.data.objects["Armature"]
 arm = obj.data
 
-# Set the keyframe at frame 1.
+# set the keyframe at frame 1
 arm.bones["Bone"].my_prop_group.nested = 10
-arm.keyframe_insert(
-    data_path='bones["Bone"].my_prop.nested',
-    frame=1,
-    group="Nested Group",
-)
+arm.keyframe_insert(data_path='bones["Bone"].my_prop.nested',
+                    frame=1,
+                    group="Nested Group")

doc/python_api/rst/bgl.rst

@@ -26,6 +26,58 @@ offers a set of extensive examples, including advanced features.
    for example.
 
 
+.. function:: glAccum(op, value):
+
+   Operate on the accumulation buffer.
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glAccum.xml>`__
+
+   :type op: Enumerated constant
+   :arg op: The accumulation buffer operation.
+   :type value: float
+   :arg value: a value used in the accumulation buffer operation.
+
+
+.. function:: glAlphaFunc(func, ref):
+
+   Specify the alpha test function.
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glAlphaFunc.xml>`__
+
+   :type func: Enumerated constant
+   :arg func: Specifies the alpha comparison function.
+   :type ref: float
+   :arg ref: The reference value that incoming alpha values are compared to.
+      Clamped between 0 and 1.
+
+
+.. function:: glAreTexturesResident(n, textures, residences):
+
+   Determine if textures are loaded in texture memory
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glAreTexturesResident.xml>`__
+
+   :type n: int
+   :arg n: Specifies the number of textures to be queried.
+   :type textures: :class:`bgl.Buffer` object I{type GL_INT}
+   :arg textures: Specifies an array containing the names of the textures to be queried
+   :type residences: :class:`bgl.Buffer` object I{type GL_INT}(boolean)
+   :arg residences: An array in which the texture residence status in returned.
+      The residence status of a texture named by an element of textures is
+      returned in the corresponding element of residences.
+
+
+.. function:: glBegin(mode):
+
+   Delimit the vertices of a primitive or a group of like primitives
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glBegin.xml>`__
+
+   :type mode: Enumerated constant
+   :arg mode: Specifies the primitive that will be create from vertices between
+      glBegin and glEnd.
+
+
 .. function:: glBindTexture(target, texture):
 
    Bind a named texture to a texturing target
@@ -38,6 +90,24 @@ offers a set of extensive examples, including advanced features.
    :arg texture: Specifies the name of a texture.
 
 
+.. function:: glBitmap(width, height, xorig, yorig, xmove, ymove, bitmap):
+
+   Draw a bitmap
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glBitmap.xml>`__
+
+   :type width, height: int
+   :arg width, height: Specify the pixel width and height of the bitmap image.
+   :type xorig, yorig: float
+   :arg xorig, yorig: Specify the location of the origin in the bitmap image. The origin is measured
+      from the lower left corner of the bitmap, with right and up being the positive axes.
+   :type xmove, ymove: float
+   :arg xmove, ymove: Specify the x and y offsets to be added to the current raster position after
+      the bitmap is drawn.
+   :type bitmap: :class:`bgl.Buffer` object I{type GL_BYTE}
+   :arg bitmap: Specifies the address of the bitmap image.
+
+
 .. function:: glBlendFunc(sfactor, dfactor):
 
    Specify pixel arithmetic
@@ -52,6 +122,32 @@ offers a set of extensive examples, including advanced features.
       blending factors are computed.
 
 
+.. function:: glCallList(list):
+
+   Execute a display list
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glCallList.xml>`__
+
+   :type list: unsigned int
+   :arg list: Specifies the integer name of the display list to be executed.
+
+
+.. function:: glCallLists(n, type, lists):
+
+   Execute a list of display lists
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glCallLists.xml>`__
+
+   :type n: int
+   :arg n: Specifies the number of display lists to be executed.
+   :type type: Enumerated constant
+   :arg type: Specifies the type of values in lists.
+   :type lists: :class:`bgl.Buffer` object
+   :arg lists: Specifies the address of an array of name offsets in the display list.
+      The pointer type is void because the offsets can be bytes, shorts, ints, or floats,
+      depending on the value of type.
+
+
 .. function:: glClear(mask):
 
    Clear buffers to preset values
@@ -62,6 +158,17 @@ offers a set of extensive examples, including advanced features.
    :arg mask: Bitwise OR of masks that indicate the buffers to be cleared.
 
 
+.. function:: glClearAccum(red, green, blue, alpha):
+
+   Specify clear values for the accumulation buffer
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glClearAccum.xml>`__
+
+   :type red, green, blue, alpha: float
+   :arg red, green, blue, alpha: Specify the red, green, blue, and alpha values used when the
+      accumulation buffer is cleared. The initial values are all 0.
+
+
 .. function:: glClearColor(red, green, blue, alpha):
 
    Specify clear values for the color buffers
@@ -84,6 +191,17 @@ offers a set of extensive examples, including advanced features.
       The initial value is 1.
 
 
+.. function:: glClearIndex(c):
+
+   Specify the clear value for the color index buffers
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glClearIndex.xml>`__
+
+   :type c: float
+   :arg c: Specifies the index used when the color index buffers are cleared.
+      The initial value is 0.
+
+
 .. function:: glClearStencil(s):
 
    Specify the clear value for the stencil buffer
@@ -137,7 +255,36 @@ offers a set of extensive examples, including advanced features.
       color components can be written.
 
 
-.. function:: glCopyTexImage2D(target, level, internalformat, x, y, width, height, border):
+.. function:: glColorMaterial(face, mode):
+
+   Cause a material color to track the current color
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glColorMaterial.xml>`__
+
+   :type face: Enumerated constant
+   :arg face: Specifies whether front, back, or both front and back material parameters should
+      track the current color.
+   :type mode: Enumerated constant
+   :arg mode: Specifies which of several material parameters track the current color.
+
+
+.. function:: glCopyPixels(x, y, width, height, type):
+
+   Copy pixels in the frame buffer
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glCopyPixels.xml>`__
+
+   :type x, y: int
+   :arg x, y: Specify the window coordinates of the lower left corner of the rectangular
+      region of pixels to be copied.
+   :type width, height: int
+   :arg width,height: Specify the dimensions of the rectangular region of pixels to be copied.
+      Both must be non-negative.
+   :type type: Enumerated constant
+   :arg type: Specifies whether color values, depth values, or stencil values are to be copied.
+
+
+   def glCopyTexImage2D(target, level, internalformat, x, y, width, height, border):
 
    Copy pixels into a 2D texture image
 
@@ -176,6 +323,18 @@ offers a set of extensive examples, including advanced features.
    :arg mode: Specifies whether front- or back-facing facets are candidates for culling.
 
 
+.. function:: glDeleteLists(list, range):
+
+   Delete a contiguous group of display lists
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glDeleteLists.xml>`__
+
+   :type list: unsigned int
+   :arg list: Specifies the integer name of the first display list to delete
+   :type range: int
+   :arg range: Specifies the number of display lists to delete
+
+
 .. function:: glDeleteTextures(n, textures):
 
    Delete named textures
@@ -244,6 +403,23 @@ offers a set of extensive examples, including advanced features.
    :arg mode: Specifies up to four color buffers to be drawn into.
 
 
+.. function:: glDrawPixels(width, height, format, type, pixels):
+
+   Write a block of pixels to the frame buffer
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glDrawPixels.xml>`__
+
+   :type width, height: int
+   :arg width, height: Specify the dimensions of the pixel rectangle to be
+      written into the frame buffer.
+   :type format: Enumerated constant
+   :arg format: Specifies the format of the pixel data.
+   :type type: Enumerated constant
+   :arg type: Specifies the data type for pixels.
+   :type pixels: :class:`bgl.Buffer` object
+   :arg pixels: Specifies a pointer to the pixel data.
+
+
 .. function:: glEdgeFlag (flag):
 
    B{glEdgeFlag, glEdgeFlagv}
@@ -266,6 +442,20 @@ offers a set of extensive examples, including advanced features.
    :arg cap: Specifies a symbolic constant indicating a GL capability.
 
 
+.. function:: glEnd():
+
+   Delimit the vertices of a primitive or group of like primitives
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glBegin.xml>`__
+
+
+.. function:: glEndList():
+
+   Create or replace a display list
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glNewList.xml>`__
+
+
 .. function:: glEvalCoord (u,v):
 
    B{glEvalCoord1d, glEvalCoord1f, glEvalCoord2d, glEvalCoord2f, glEvalCoord1dv, glEvalCoord1fv,
@@ -370,6 +560,33 @@ offers a set of extensive examples, including advanced features.
    :arg mode: Specifies the orientation of front-facing polygons.
 
 
+.. function:: glFrustum(left, right, bottom, top, zNear, zFar):
+
+   Multiply the current matrix by a perspective matrix
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glFrustum.xml>`__
+
+   :type left, right: double (float)
+   :arg left, right: Specify the coordinates for the left and right vertical
+      clipping planes.
+   :type top, bottom: double (float)
+   :arg top, bottom: Specify the coordinates for the bottom and top horizontal
+      clipping planes.
+   :type zNear, zFar: double (float)
+   :arg zNear, zFar: Specify the distances to the near and far depth clipping planes.
+      Both distances must be positive.
+
+
+.. function:: glGenLists(range):
+
+   Generate a contiguous set of empty display lists
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glGenLists.xml>`__
+
+   :type range: int
+   :arg range: Specifies the number of contiguous empty display lists to be generated.
+
+
 .. function:: glGenTextures(n, textures):
 
    Generate texture names
@@ -396,6 +613,21 @@ offers a set of extensive examples, including advanced features.
    :arg param: Returns the value or values of the specified parameter.
 
 
+.. function:: glGetClipPlane(plane, equation):
+
+   Return the coefficients of the specified clipping plane
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glGetClipPlane.xml>`__
+
+   :type plane: Enumerated constant
+   :arg plane: Specifies a clipping plane. The number of clipping planes depends on the
+      implementation, but at least six clipping planes are supported. They are identified by
+      symbolic names of the form GL_CLIP_PLANEi where 0 < i < GL_MAX_CLIP_PLANES.
+   :type equation:  :class:`bgl.Buffer` object I{type GL_FLOAT}
+   :arg equation:  Returns four float (double)-precision values that are the coefficients of the
+      plane equation of plane in eye coordinates. The initial value is (0, 0, 0, 0).
+
+
 .. function:: glGetError():
 
    Return error information
@@ -468,6 +700,16 @@ offers a set of extensive examples, including advanced features.
    :arg values: Returns the pixel map contents.
 
 
+.. function:: glGetPolygonStipple(mask):
+
+   Return the polygon stipple pattern
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glGetPolygonStipple.xml>`__
+
+   :type mask: :class:`bgl.Buffer` object I{type GL_BYTE}
+   :arg mask: Returns the stipple pattern. The initial value is all 1's.
+
+
 .. function:: glGetString(name):
 
    Return a string describing the current GL connection
@@ -579,6 +821,38 @@ offers a set of extensive examples, including advanced features.
    :arg mode: Specifies a symbolic constant indicating the desired behavior.
 
 
+.. function:: glIndex(c):
+
+   B{glIndexd, glIndexf, glIndexi, glIndexs,  glIndexdv, glIndexfv, glIndexiv, glIndexsv}
+
+   Set the current color index
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glIndex.xml>`__
+
+   :type c: :class:`bgl.Buffer` object. Depends on function prototype.
+   :arg c: Specifies a pointer to a one element array that contains the new value for
+      the current color index.
+
+
+.. function:: glIndexMask(mask):
+
+   Control the writing of individual bits in the color index buffers
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glIndexMask.xml>`__
+
+   :type mask: int
+   :arg mask: Specifies a bit mask to enable and disable the writing of individual bits
+      in the color index buffers.
+      Initially, the mask is all 1's.
+
+
+.. function:: glInitNames():
+
+   Initialize the name stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glInitNames.xml>`__
+
+
 .. function:: glIsEnabled(cap):
 
    Test whether a capability is enabled
@@ -589,6 +863,16 @@ offers a set of extensive examples, including advanced features.
    :arg cap: Specifies a constant representing a GL capability.
 
 
+.. function:: glIsList(list):
+
+   Determine if a name corresponds to a display-list
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glIsList.xml>`__
+
+   :type list: unsigned int
+   :arg list: Specifies a potential display-list name.
+
+
 .. function:: glIsTexture(texture):
 
    Determine if a name corresponds to a texture
@@ -634,6 +918,23 @@ offers a set of extensive examples, including advanced features.
       specifies a pointer to the value or values that param will be set to.
 
 
+.. function:: glLineStipple(factor, pattern):
+
+   Specify the line stipple pattern
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glLineStipple.xml>`__
+
+   :type factor: int
+   :arg factor: Specifies a multiplier for each bit in the line stipple pattern.
+      If factor is 3, for example, each bit in the pattern is used three times before
+      the next bit in the pattern is used. factor is clamped to the range [1, 256] and
+      defaults to 1.
+   :type pattern: unsigned short int
+   :arg pattern: Specifies a 16-bit integer whose bit pattern determines which fragments
+      of a line will be drawn when the line is rasterized. Bit zero is used first; the default
+      pattern is all 1's.
+
+
 .. function:: glLineWidth(width):
 
    Specify the width of rasterized lines.
@@ -644,6 +945,24 @@ offers a set of extensive examples, including advanced features.
    :arg width: Specifies the width of rasterized lines. The initial value is 1.
 
 
+.. function:: glListBase(base):
+
+   Set the display-list base for glCallLists
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glListBase.xml>`__
+
+   :type base: unsigned int
+   :arg base: Specifies an integer offset that will be added to glCallLists
+      offsets to generate display-list names. The initial value is 0.
+
+
+.. function:: glLoadIdentity():
+
+   Replace the current matrix with the identity matrix
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glLoadIdentity.xml>`__
+
+
 .. function:: glLoadMatrix (m):
 
    B{glLoadMatrixd, glLoadMatixf}
@@ -657,6 +976,16 @@ offers a set of extensive examples, including advanced features.
       of a 4x4 column-major matrix.
 
 
+.. function:: glLoadName(name):
+
+   Load a name onto the name stack.
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glLoadName.xml>`__
+
+   :type name: unsigned int
+   :arg name: Specifies a name that will replace the top value on the name stack.
+
+
 .. function:: glLogicOp(opcode):
 
    Specify a logical pixel operation for color index rendering
@@ -771,6 +1100,16 @@ offers a set of extensive examples, including advanced features.
       pname will be set to.
 
 
+.. function:: glMatrixMode(mode):
+
+   Specify which matrix is the current matrix.
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glMatrixMode.xml>`__
+
+   :type mode: Enumerated constant
+   :arg mode: Specifies which matrix stack is the target for subsequent matrix operations.
+
+
 .. function:: glMultMatrix (m):
 
    B{glMultMatrixd, glMultMatrixf}
@@ -784,6 +1123,18 @@ offers a set of extensive examples, including advanced features.
       major matrix.
 
 
+.. function:: glNewList(list, mode):
+
+   Create or replace a display list
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glNewList.xml>`__
+
+   :type list: unsigned int
+   :arg list: Specifies the display list name
+   :type mode: Enumerated constant
+   :arg mode: Specifies the compilation mode.
+
+
 .. function:: glNormal3 (nx, ny, nz, v):
 
    B{Normal3b, Normal3bv, Normal3d, Normal3dv, Normal3f, Normal3fv, Normal3i, Normal3iv,
@@ -801,6 +1152,34 @@ offers a set of extensive examples, including advanced features.
       of the new current normal.
 
 
+.. function:: glOrtho(left, right, bottom, top, zNear, zFar):
+
+   Multiply the current matrix with an orthographic matrix
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glOrtho.xml>`__
+
+   :type left, right: double (float)
+   :arg left, right: Specify the coordinates for the left and
+      right vertical clipping planes.
+   :type bottom, top: double (float)
+   :arg bottom, top: Specify the coordinates for the bottom and top
+      horizontal clipping planes.
+   :type zNear, zFar: double (float)
+   :arg zNear, zFar: Specify the distances to the nearer and farther
+      depth clipping planes. These values are negative if the plane is to be behind the viewer.
+
+
+.. function:: glPassThrough(token):
+
+   Place a marker in the feedback buffer
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPassThrough.xml>`__
+
+   :type token: float
+   :arg token: Specifies a marker value to be placed in the feedback
+      buffer following a GL_PASS_THROUGH_TOKEN.
+
+
 .. function:: glPixelMap (map, mapsize, values):
 
    B{glPixelMapfv, glPixelMapuiv, glPixelMapusv}
@@ -847,6 +1226,16 @@ offers a set of extensive examples, including advanced features.
    :arg param: Specifies the value that pname is set to.
 
 
+.. function:: glPixelZoom(xfactor, yfactor):
+
+   Specify the pixel zoom factors
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPixelZoom.xml>`__
+
+   :type xfactor, yfactor: float
+   :arg xfactor, yfactor: Specify the x and y zoom factors for pixel write operations.
+
+
 .. function:: glPointSize(size):
 
    Specify the diameter of rasterized points
@@ -886,6 +1275,98 @@ offers a set of extensive examples, including advanced features.
       constant depth offset. The initial value is 0.
 
 
+.. function:: glPolygonStipple(mask):
+
+   Set the polygon stippling pattern
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPolygonStipple.xml>`__
+
+   :type mask: :class:`bgl.Buffer` object I{type GL_BYTE}
+   :arg mask: Specifies a pointer to a 32x32 stipple pattern that will be unpacked
+      from memory in the same way that glDrawPixels unpacks pixels.
+
+
+.. function:: glPopAttrib():
+
+   Pop the server attribute stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPopAttrib.xml>`__
+
+
+.. function:: glPopClientAttrib():
+
+   Pop the client attribute stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPopClientAttrib.xml>`__
+
+
+.. function:: glPopMatrix():
+
+   Pop the current matrix stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPopMatrix.xml>`__
+
+
+.. function:: glPopName():
+
+   Pop the name stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPopName.xml>`__
+
+
+.. function:: glPrioritizeTextures(n, textures, priorities):
+
+   Set texture residence priority
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPrioritizeTextures.xml>`__
+
+   :type n: int
+   :arg n: Specifies the number of textures to be prioritized.
+   :type textures: :class:`bgl.Buffer` I{type GL_INT}
+   :arg textures: Specifies an array containing the names of the textures to be prioritized.
+   :type priorities: :class:`bgl.Buffer` I{type GL_FLOAT}
+   :arg priorities: Specifies an array containing the texture priorities.
+      A priority given in an element of priorities applies to the texture named
+      by the corresponding element of textures.
+
+
+.. function:: glPushAttrib(mask):
+
+   Push the server attribute stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPushAttrib.xml>`__
+
+   :type mask: Enumerated constant(s)
+   :arg mask: Specifies a mask that indicates which attributes to save.
+
+
+.. function:: glPushClientAttrib(mask):
+
+   Push the client attribute stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPushClientAttrib.xml>`__
+
+   :type mask: Enumerated constant(s)
+   :arg mask: Specifies a mask that indicates which attributes to save.
+
+
+.. function:: glPushMatrix():
+
+   Push the current matrix stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPushMatrix.xml>`__
+
+
+.. function:: glPushName(name):
+
+   Push the name stack
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glPushName.xml>`__
+
+   :type name: unsigned int
+   :arg name: Specifies a name that will be pushed onto the name stack.
+
+
 .. function:: glRasterPos (x,y,z,w):
 
    B{glRasterPos2d, glRasterPos2f, glRasterPos2i, glRasterPos2s, glRasterPos3d,
@@ -972,6 +1453,16 @@ offers a set of extensive examples, including advanced features.
       to the opposite vertex of the rectangle
 
 
+.. function:: glRenderMode(mode):
+
+   Set rasterization mode
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glRenderMode.xml>`__
+
+   :type mode: Enumerated constant
+   :arg mode: Specifies the rasterization mode.
+
+
 .. function:: glRotate (angle, x, y, z):
 
    B{glRotated, glRotatef}
@@ -1012,6 +1503,28 @@ offers a set of extensive examples, including advanced features.
       dimensions of that window.
 
 
+.. function:: glSelectBuffer(size, buffer):
+
+   Establish a buffer for selection mode values
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glSelectBuffer.xml>`__
+
+   :type size: int
+   :arg size: Specifies the size of buffer
+   :type buffer: :class:`bgl.Buffer` I{type GL_INT}
+   :arg buffer: Returns the selection data
+
+
+.. function:: glShadeModel(mode):
+
+   Select flat or smooth shading
+
+   .. seealso:: `OpenGL Docs <https://www.opengl.org/sdk/docs/man2/xhtml/glShadeModel.xml>`__
+
+   :type mode: Enumerated constant
+   :arg mode: Specifies a symbolic value representing a shading technique.
+
+
 .. function:: glStencilFunc(func, ref, mask):
 
    Set function and reference value for stencil testing
@@ -1244,6 +1757,99 @@ offers a set of extensive examples, including advanced features.
       dimensions of that window.
 
 
+.. function:: gluPerspective(fovY, aspect, zNear, zFar):
+
+   Set up a perspective projection matrix.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluPerspective.xml
+
+   :type fovY: double
+   :arg fovY: Specifies the field of view angle, in degrees, in the y direction.
+   :type aspect: double
+   :arg aspect: Specifies the aspect ratio that determines the field of view in the x direction.
+    The aspect ratio is the ratio of x (width) to y (height).
+   :type zNear: double
+   :arg zNear: Specifies the distance from the viewer to the near clipping plane (always positive).
+   :type zFar: double
+   :arg zFar: Specifies the distance from the viewer to the far clipping plane (always positive).
+
+
+.. function:: gluLookAt(eyex, eyey, eyez, centerx, centery, centerz, upx, upy, upz):
+
+   Define a viewing transformation.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluLookAt.xml
+
+   :type eyex, eyey, eyez: double
+   :arg eyex, eyey, eyez: Specifies the position of the eye point.
+   :type centerx, centery, centerz: double
+   :arg centerx, centery, centerz: Specifies the position of the reference point.
+   :type upx, upy, upz: double
+   :arg upx, upy, upz: Specifies the direction of the up vector.
+
+
+.. function:: gluOrtho2D(left, right, bottom, top):
+
+   Define a 2-D orthographic projection matrix.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluOrtho2D.xml
+
+   :type left, right: double
+   :arg left, right: Specify the coordinates for the left and right vertical clipping planes.
+   :type bottom, top: double
+   :arg bottom, top: Specify the coordinates for the bottom and top horizontal clipping planes.
+
+
+.. function:: gluPickMatrix(x, y, width, height, viewport):
+
+   Define a picking region.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluPickMatrix.xml
+
+   :type x, y: double
+   :arg x, y: Specify the center of a picking region in window coordinates.
+   :type width, height: double
+   :arg width, height: Specify the width and height, respectively, of the picking region in window coordinates.
+   :type viewport: :class:`bgl.Buffer` object. [int]
+   :arg viewport: Specifies the current viewport.
+
+
+.. function:: gluProject(objx, objy, objz, modelMatrix, projMatrix, viewport, winx, winy, winz):
+
+   Map object coordinates to window coordinates.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluProject.xml
+
+   :type objx, objy, objz: double
+   :arg objx, objy, objz: Specify the object coordinates.
+   :type modelMatrix: :class:`bgl.Buffer` object. [double]
+   :arg modelMatrix: Specifies the current modelview matrix (as from a glGetDoublev call).
+   :type projMatrix: :class:`bgl.Buffer` object. [double]
+   :arg projMatrix: Specifies the current projection matrix (as from a glGetDoublev call).
+   :type viewport: :class:`bgl.Buffer` object. [int]
+   :arg viewport: Specifies the current viewport (as from a glGetIntegerv call).
+   :type winx, winy, winz: :class:`bgl.Buffer` object. [double]
+   :arg winx, winy, winz: Return the computed window coordinates.
+
+
+.. function:: gluUnProject(winx, winy, winz, modelMatrix, projMatrix, viewport, objx, objy, objz):
+
+   Map object coordinates to window coordinates.
+
+   .. seealso:: https://www.opengl.org/sdk/docs/man2/xhtml/gluUnProject.xml
+
+   :type winx, winy, winz: double
+   :arg winx, winy, winz: Specify the window coordinates to be mapped.
+   :type modelMatrix: :class:`bgl.Buffer` object. [double]
+   :arg modelMatrix: Specifies the current modelview matrix (as from a glGetDoublev call).
+   :type projMatrix: :class:`bgl.Buffer` object. [double]
+   :arg projMatrix: Specifies the current projection matrix (as from a glGetDoublev call).
+   :type viewport: :class:`bgl.Buffer` object. [int]
+   :arg viewport: Specifies the current viewport (as from a glGetIntegerv call).
+   :type objx, objy, objz: :class:`bgl.Buffer` object. [double]
+   :arg objx, objy, objz: Return the computed object coordinates.
+
+
 .. function:: glUseProgram(program):
 
    Installs a program object as part of current rendering state

doc/python_api/rst/info_gotcha.rst

@@ -633,7 +633,6 @@ Here are some general hints to avoid running into these problems.
   fetch data from the context each time the script is activated.
 - Crashes may not happen every time, they may happen more on some configurations/operating-systems.
 - Be wary of recursive patterns, those are very efficient at hiding the issues described here.
-- See last sub-section about `Unfortunate Corner Cases`_ for some known breaking exceptions.
 
 .. note::
 
@@ -828,17 +827,6 @@ the next example will still crash.
    print(vertices)  # <- this may crash
 
 
-Unfortunate Corner Cases
-------------------------
-
-Besides all expected cases listed above, there are a few others that should not be
-an issue but, due to internal implementation details, currently are:
-
-- ``Object.hide_viewport``, ``Object.hide_select`` and ``Object.hide_render``:
-  Setting any of those booleans will trigger a rebuild of Collection caches, hence breaking
-  any current iteration over ``Collection.all_objects``.
-
-
 sys.exit
 ========
 

doc/python_api/sphinx_doc_gen.py

@@ -1009,8 +1009,7 @@ context_type_map = {
     "editable_gpencil_strokes": ("GPencilStroke", True),
     "editable_objects": ("Object", True),
     "fluid": ("FluidSimulationModifier", False),
-    "gpencil": ("GreasePencil", False),
-    "gpencil_data": ("GreasePencil", False),
+    "gpencil_data": ("GreasePencel", False),
     "gpencil_data_owner": ("ID", False),
     "image_paint_object": ("Object", False),
     "lattice": ("Lattice", False),
@@ -1029,7 +1028,6 @@ context_type_map = {
     "particle_system": ("ParticleSystem", False),
     "particle_system_editable": ("ParticleSystem", False),
     "pose_bone": ("PoseBone", False),
-    "pose_object": ("Object", False),
     "scene": ("Scene", False),
     "sculpt_object": ("Object", False),
     "selectable_bases": ("ObjectBase", True),

extern/audaspace/CMakeLists.txt

@@ -73,8 +73,6 @@ set(SRC
 	src/fx/LoopReader.cpp
 	src/fx/LowpassCalculator.cpp
 	src/fx/Lowpass.cpp
-	src/fx/Modulator.cpp
-	src/fx/ModulatorReader.cpp
 	src/fx/MutableReader.cpp
 	src/fx/MutableSound.cpp
 	src/fx/Pitch.cpp
@@ -183,8 +181,6 @@ set(PUBLIC_HDR
 	include/fx/LoopReader.h
 	include/fx/LowpassCalculator.h
 	include/fx/Lowpass.h
-	include/fx/Modulator.h
-	include/fx/ModulatorReader.h
 	include/fx/MutableReader.h
 	include/fx/MutableSound.h
 	include/fx/Pitch.h

extern/audaspace/bindings/C/AUD_Sound.cpp

@@ -32,7 +32,6 @@
 #include "fx/Limiter.h"
 #include "fx/Loop.h"
 #include "fx/Lowpass.h"
-#include "fx/Modulator.h"
 #include "fx/Pitch.h"
 #include "fx/Reverse.h"
 #include "fx/Sum.h"
@@ -466,21 +465,6 @@ AUD_API AUD_Sound* AUD_Sound_lowpass(AUD_Sound* sound, float frequency, float Q)
 	}
 }
 
-AUD_API AUD_Sound* AUD_Sound_modulate(AUD_Sound* first, AUD_Sound* second)
-{
-	assert(first);
-	assert(second);
-
-	try
-	{
-		return new AUD_Sound(new Modulator(*first, *second));
-	}
-	catch(Exception&)
-	{
-		return nullptr;
-	}
-}
-
 AUD_API AUD_Sound* AUD_Sound_pitch(AUD_Sound* sound, float factor)
 {
 	assert(sound);

extern/audaspace/bindings/C/AUD_Sound.h

@@ -246,14 +246,6 @@ extern AUD_API AUD_Sound* AUD_Sound_loop(AUD_Sound* sound, int count);
  */
 extern AUD_API AUD_Sound* AUD_Sound_lowpass(AUD_Sound* sound, float frequency, float Q);
 
-/**
- * Modulates two sound, which means multiplying the sound samples.
- * \param first The first sound.
- * \param second The second sound.
- * \return A handle of the modulated sound.
- */
-extern AUD_API AUD_Sound* AUD_Sound_modulate(AUD_Sound* first, AUD_Sound* second);
-
 /**
  * Changes the pitch of a sound.
  * \param sound The sound to change.

extern/audaspace/bindings/python/PySound.cpp

@@ -42,7 +42,6 @@
 #include "fx/Limiter.h"
 #include "fx/Loop.h"
 #include "fx/Lowpass.h"
-#include "fx/Modulator.h"
 #include "fx/MutableSound.h"
 #include "fx/Pitch.h"
 #include "fx/Reverse.h"
@@ -1130,47 +1129,6 @@ Sound_lowpass(Sound* self, PyObject* args)
 	return (PyObject *)parent;
 }
 
-PyDoc_STRVAR(M_aud_Sound_modulate_doc,
-			 "modulate(sound)\n\n"
-			 "Modulates two factories.\n\n"
-			 ":arg sound: The sound to modulate over the other.\n"
-			 ":type sound: :class:`Sound`\n"
-			 ":return: The created :class:`Sound` object.\n"
-			 ":rtype: :class:`Sound`\n\n"
-			 ".. note:: The two factories have to have the same specifications "
-			 "(channels and samplerate).");
-
-static PyObject *
-Sound_modulate(Sound* self, PyObject* object)
-{
-	PyTypeObject* type = Py_TYPE(self);
-
-	if(!PyObject_TypeCheck(object, type))
-	{
-		PyErr_SetString(PyExc_TypeError, "Object is not of type Sound!");
-		return nullptr;
-	}
-
-	Sound* parent = (Sound*)type->tp_alloc(type, 0);
-	Sound* child = (Sound*)object;
-
-	if(parent != nullptr)
-	{
-		try
-		{
-			parent->sound = new std::shared_ptr<ISound>(new Modulator(*reinterpret_cast<std::shared_ptr<ISound>*>(self->sound), *reinterpret_cast<std::shared_ptr<ISound>*>(child->sound)));
-		}
-		catch(Exception& e)
-		{
-			Py_DECREF(parent);
-			PyErr_SetString(AUDError, e.what());
-			return nullptr;
-		}
-	}
-
-	return (PyObject *)parent;
-}
-
 PyDoc_STRVAR(M_aud_Sound_pitch_doc,
 			 "pitch(factor)\n\n"
 			 "Changes the pitch of a sound with a specific factor.\n\n"
@@ -1833,9 +1791,6 @@ static PyMethodDef Sound_methods[] = {
 	{"lowpass", (PyCFunction)Sound_lowpass, METH_VARARGS,
 	 M_aud_Sound_lowpass_doc
 	},
-	{"modulate", (PyCFunction)Sound_modulate, METH_O,
-	 M_aud_Sound_modulate_doc
-	},
 	{"pitch", (PyCFunction)Sound_pitch, METH_VARARGS,
 	 M_aud_Sound_pitch_doc
 	},

extern/audaspace/include/fx/Modulator.h

@@ -1,62 +0,0 @@
-/*******************************************************************************
- * Copyright 2009-2016 Jörg Müller
- *
- * 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 Modulator.h
- * @ingroup fx
- * The Modulator class.
- */
-
-#include "ISound.h"
-
-AUD_NAMESPACE_BEGIN
-
-/**
- * This sound plays two other factories, playing them the same time and modulating/multiplying them.
- * \note Readers from the underlying factories must have the same sample rate
- *       and channel count.
- */
-class AUD_API Modulator : public ISound
-{
-private:
-	/**
-	 * First played sound.
-	 */
-	std::shared_ptr<ISound> m_sound1;
-
-	/**
-	 * Second played sound.
-	 */
-	std::shared_ptr<ISound> m_sound2;
-
-	// delete copy constructor and operator=
-	Modulator(const Modulator&) = delete;
-	Modulator& operator=(const Modulator&) = delete;
-
-public:
-	/**
-	 * Creates a new modulator sound.
-	 * \param sound1 The first input sound.
-	 * \param sound2 The second input sound.
-	 */
-	Modulator(std::shared_ptr<ISound> sound1, std::shared_ptr<ISound> sound2);
-
-	virtual std::shared_ptr<IReader> createReader();
-};
-
-AUD_NAMESPACE_END

extern/audaspace/include/fx/ModulatorReader.h

@@ -1,79 +0,0 @@
-/*******************************************************************************
- * Copyright 2009-2016 Jörg Müller
- *
- * 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 ModulatorReader.h
- * @ingroup fx
- * The ModulatorReader class.
- */
-
-#include "IReader.h"
-#include "util/Buffer.h"
-
-#include <memory>
-
-AUD_NAMESPACE_BEGIN
-
-/**
- * This reader plays two readers with the same specs in parallel multiplying their samples.
- */
-class AUD_API ModulatorReader : public IReader
-{
-private:
-	/**
-	 * The first reader.
-	 */
-	std::shared_ptr<IReader> m_reader1;
-
-	/**
-	 * The second reader.
-	 */
-	std::shared_ptr<IReader> m_reader2;
-
-	/**
-	 * Buffer used for mixing.
-	 */
-	Buffer m_buffer;
-
-	// delete copy constructor and operator=
-	ModulatorReader(const ModulatorReader&) = delete;
-	ModulatorReader& operator=(const ModulatorReader&) = delete;
-
-public:
-	/**
-	 * Creates a new modulator reader.
-	 * \param reader1 The first reader to read from.
-	 * \param reader2 The second reader to read from.
-	 * \exception Exception Thrown if the specs from the readers differ.
-	 */
-	ModulatorReader(std::shared_ptr<IReader> reader1, std::shared_ptr<IReader> reader2);
-
-	/**
-	 * Destroys the reader.
-	 */
-	virtual ~ModulatorReader();
-
-	virtual bool isSeekable() const;
-	virtual void seek(int position);
-	virtual int getLength() const;
-	virtual int getPosition() const;
-	virtual Specs getSpecs() const;
-	virtual void read(int& length, bool& eos, sample_t* buffer);
-};
-
-AUD_NAMESPACE_END

extern/audaspace/src/devices/DeviceManager.cpp

@@ -35,12 +35,7 @@ void DeviceManager::registerDevice(std::string name, std::shared_ptr<IDeviceFact
 
 std::shared_ptr<IDeviceFactory> DeviceManager::getDeviceFactory(std::string name)
 {
-	auto it = m_factories.find(name);
-
-	if(it == m_factories.end())
-		return nullptr;
-
-	return it->second;
+	return m_factories[name];
 }
 
 std::shared_ptr<IDeviceFactory> DeviceManager::getDefaultDeviceFactory()

extern/audaspace/src/fx/Modulator.cpp

@@ -1,35 +0,0 @@
-/*******************************************************************************
- * Copyright 2009-2016 Jörg Müller
- *
- * 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/Modulator.h"
-#include "fx/ModulatorReader.h"
-
-AUD_NAMESPACE_BEGIN
-
-Modulator::Modulator(std::shared_ptr<ISound> sound1, std::shared_ptr<ISound> sound2) :
-	m_sound1(sound1), m_sound2(sound2)
-{
-}
-
-std::shared_ptr<IReader> Modulator::createReader()
-{
-	std::shared_ptr<IReader> reader1 = m_sound1->createReader();
-	std::shared_ptr<IReader> reader2 = m_sound2->createReader();
-
-	return std::shared_ptr<IReader>(new ModulatorReader(reader1, reader2));
-}
-
-AUD_NAMESPACE_END

extern/audaspace/src/fx/ModulatorReader.cpp

@@ -1,95 +0,0 @@
-/*******************************************************************************
- * Copyright 2009-2016 Jörg Müller
- *
- * 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/ModulatorReader.h"
-#include "Exception.h"
-
-#include <algorithm>
-#include <cstring>
-
-AUD_NAMESPACE_BEGIN
-
-ModulatorReader::ModulatorReader(std::shared_ptr<IReader> reader1, std::shared_ptr<IReader> reader2) :
-    m_reader1(reader1), m_reader2(reader2)
-{
-}
-
-ModulatorReader::~ModulatorReader()
-{
-}
-
-bool ModulatorReader::isSeekable() const
-{
-	return m_reader1->isSeekable() && m_reader2->isSeekable();
-}
-
-void ModulatorReader::seek(int position)
-{
-	m_reader1->seek(position);
-	m_reader2->seek(position);
-}
-
-int ModulatorReader::getLength() const
-{
-	int len1 = m_reader1->getLength();
-	int len2 = m_reader2->getLength();
-	if((len1 < 0) || (len2 < 0))
-		return -1;
-	return std::min(len1, len2);
-}
-
-int ModulatorReader::getPosition() const
-{
-	int pos1 = m_reader1->getPosition();
-	int pos2 = m_reader2->getPosition();
-	return std::max(pos1, pos2);
-}
-
-Specs ModulatorReader::getSpecs() const
-{
-	return m_reader1->getSpecs();
-}
-
-void ModulatorReader::read(int& length, bool& eos, sample_t* buffer)
-{
-	Specs specs = m_reader1->getSpecs();
-	Specs s2 = m_reader2->getSpecs();
-	if(!AUD_COMPARE_SPECS(specs, s2))
-		AUD_THROW(StateException, "Two readers with different specifiactions cannot be modulated.");
-
-	int samplesize = AUD_SAMPLE_SIZE(specs);
-
-	m_buffer.assureSize(length * samplesize);
-
-	int len1 = length;
-	m_reader1->read(len1, eos, buffer);
-
-	if(len1 < length)
-		std::memset(buffer + len1 * specs.channels, 0, (length - len1) * samplesize);
-
-	int len2 = length;
-	bool eos2;
-	sample_t* buf = m_buffer.getBuffer();
-	m_reader2->read(len2, eos2, buf);
-
-	for(int i = 0; i < len2 * specs.channels; i++)
-		buffer[i] *= buf[i];
-
-	length = std::max(len1, len2);
-	eos &= eos2;
-}
-
-AUD_NAMESPACE_END

intern/audaspace/intern/AUD_PyInit.cpp

@@ -66,11 +66,6 @@ static PyMethodDef meth_sound_from_pointer[] = {
 PyObject *AUD_initPython(void)
 {
 	PyObject *module = PyInit_aud();
-	if (module == NULL) {
-		printf("Unable to initialise audio\n");
-		return NULL;
-	}
-
 	PyModule_AddObject(module, "_sound_from_pointer", (PyObject *)PyCFunction_New(meth_sound_from_pointer, NULL));
 	PyDict_SetItemString(PyImport_GetModuleDict(), "aud", module);
 

intern/cycles/CMakeLists.txt

@@ -301,8 +301,6 @@ if(WITH_CYCLES_CUDA_BINARIES AND (NOT WITH_CYCLES_CUBIN_COMPILER))
 			set(MAX_MSVC 1911)
 		elseif(${CUDA_VERSION} EQUAL "10.0")
 			set(MAX_MSVC 1999)
-		elseif(${CUDA_VERSION} EQUAL "10.1")
-			set(MAX_MSVC 1999)
 		endif()
 		if(NOT MSVC_VERSION LESS ${MAX_MSVC} OR CMAKE_C_COMPILER_ID MATCHES "Clang")
 			message(STATUS "nvcc not supported for this compiler version, using cycles_cubin_cc instead.")

intern/cycles/app/CMakeLists.txt

@@ -58,8 +58,7 @@ link_directories(${OPENIMAGEIO_LIBPATH}
                  ${JPEG_LIBPATH}
                  ${ZLIB_LIBPATH}
                  ${TIFF_LIBPATH}
-                 ${OPENEXR_LIBPATH}
-                 ${OPENJPEG_LIBPATH})
+                 ${OPENEXR_LIBPATH})
 
 if(WITH_OPENCOLORIO)
 	link_directories(${OPENCOLORIO_LIBPATH})
@@ -95,7 +94,6 @@ macro(cycles_target_link_libraries target)
 		${target}
 		${OPENIMAGEIO_LIBRARIES}
 		${OPENEXR_LIBRARIES}
-		${OPENJPEG_LIBRARIES}
 		${PUGIXML_LIBRARIES}
 		${BOOST_LIBRARIES}
 		${CMAKE_DL_LIBS}
@@ -153,7 +151,6 @@ if(WITH_CYCLES_CUBIN_COMPILER)
 			extern_cuew
 			${OPENIMAGEIO_LIBRARIES}
 			${OPENEXR_LIBRARIES}
-			${OPENJPEG_LIBRARIES}
 			${PUGIXML_LIBRARIES}
 			${BOOST_LIBRARIES}
 			${PLATFORM_LINKLIBS}

intern/cycles/app/cycles_xml.cpp

@@ -442,7 +442,7 @@ static void xml_read_mesh(const XMLReadState& state, xml_node node)
 		if(xml_read_float_array(UV, node, "UV")) {
 			ustring name = ustring("UVMap");
 			Attribute *attr = mesh->attributes.add(ATTR_STD_UV, name);
-			float2 *fdata = attr->data_float2();
+			float3 *fdata = attr->data_float3();
 
 			/* loop over the triangles */
 			index_offset = 0;
@@ -456,9 +456,9 @@ static void xml_read_mesh(const XMLReadState& state, xml_node node)
 					assert(v1*2+1 < (int)UV.size());
 					assert(v2*2+1 < (int)UV.size());
 
-					fdata[0] = make_float2(UV[v0*2], UV[v0*2+1]);
-					fdata[1] = make_float2(UV[v1*2], UV[v1*2+1]);
-					fdata[2] = make_float2(UV[v2*2], UV[v2*2+1]);
+					fdata[0] = make_float3(UV[v0*2], UV[v0*2+1], 0.0);
+					fdata[1] = make_float3(UV[v1*2], UV[v1*2+1], 0.0);
+					fdata[2] = make_float3(UV[v2*2], UV[v2*2+1], 0.0);
 					fdata += 3;
 				}
 
@@ -516,6 +516,8 @@ static void xml_read_mesh(const XMLReadState& state, xml_node node)
 		xml_read_float(&sdparams.dicing_rate, node, "dicing_rate");
 		sdparams.dicing_rate = std::max(0.1f, sdparams.dicing_rate);
 
+		state.scene->camera->update(state.scene);
+		sdparams.camera = state.scene->camera;
 		sdparams.objecttoworld = state.tfm;
 	}
 

intern/cycles/app/io_export_cycles_xml.py

@@ -16,6 +16,7 @@
 
 # XML exporter for generating test files, not intended for end users
 
+import os
 import xml.etree.ElementTree as etree
 import xml.dom.minidom as dom
 

intern/cycles/blender/addon/operators.py

@@ -124,48 +124,9 @@ class CYCLES_OT_denoise_animation(Operator):
         return {'FINISHED'}
 
 
-class CYCLES_OT_merge_images(Operator):
-    "Combine OpenEXR multilayer images rendered with different sample" \
-    "ranges into one image with reduced noise"
-    bl_idname = "cycles.merge_images"
-    bl_label = "Merge Images"
-
-    input_filepath1: StringProperty(
-        name='Input Filepath',
-        description='File path for image to merge',
-        default='',
-        subtype='FILE_PATH')
-
-    input_filepath2: StringProperty(
-        name='Input Filepath',
-        description='File path for image to merge',
-        default='',
-        subtype='FILE_PATH')
-
-    output_filepath: StringProperty(
-        name='Output Filepath',
-        description='File path for merged image',
-        default='',
-        subtype='FILE_PATH')
-
-    def execute(self, context):
-        in_filepaths = [self.input_filepath1, self.input_filepath2]
-        out_filepath = self.output_filepath
-
-        import _cycles
-        try:
-            _cycles.merge(input=in_filepaths, output=out_filepath)
-        except Exception as e:
-            self.report({'ERROR'}, str(e))
-            return {'FINISHED'}
-
-        return {'FINISHED'}
-
-
 classes = (
     CYCLES_OT_use_shading_nodes,
-    CYCLES_OT_denoise_animation,
-    CYCLES_OT_merge_images
+    CYCLES_OT_denoise_animation
 )
 
 def register():

intern/cycles/blender/addon/properties.py

@@ -192,13 +192,13 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
     samples: IntProperty(
         name="Samples",
         description="Number of samples to render for each pixel",
-        min=1, max=(1 << 24),
+        min=1, max=2147483647,
         default=128,
     )
     preview_samples: IntProperty(
         name="Preview Samples",
         description="Number of samples to render in the viewport, unlimited if 0",
-        min=0, max=(1 << 24),
+        min=0, max=2147483647,
         default=32,
     )
     preview_pause: BoolProperty(
@@ -722,6 +722,11 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
         update=_devices_update_callback
     )
 
+    debug_opencl_kernel_single_program: BoolProperty(
+        name="Single Program",
+        default=False,
+        update=_devices_update_callback,
+    )
     del _devices_update_callback
 
     debug_use_opencl_debug: BoolProperty(name="Debug OpenCL", default=False)
@@ -1453,11 +1458,10 @@ class CyclesPreferences(bpy.types.AddonPreferences):
                 # Update name in case it changed
                 entry.name = device[0]
 
-    # Gets all devices types by default.
-    def get_devices(self, compute_device_type=''):
+    def get_devices(self):
         import _cycles
         # Layout of the device tuples: (Name, Type, Persistent ID)
-        device_list = _cycles.available_devices(compute_device_type)
+        device_list = _cycles.available_devices(self.compute_device_type)
         # Make sure device entries are up to date and not referenced before
         # we know we don't add new devices. This way we guarantee to not
         # hold pointers to a resized array.
@@ -1514,7 +1518,7 @@ class CyclesPreferences(bpy.types.AddonPreferences):
         row = layout.row()
         row.prop(self, "compute_device_type", expand=True)
 
-        cuda_devices, opencl_devices = self.get_devices(self.compute_device_type)
+        cuda_devices, opencl_devices = self.get_devices()
         row = layout.row()
         if self.compute_device_type == 'CUDA':
             self._draw_devices(row, 'CUDA', cuda_devices)

intern/cycles/blender/addon/ui.py

@@ -18,12 +18,12 @@
 
 import bpy
 from bpy_extras.node_utils import find_node_input
-from bl_ui.utils import PresetPanel
+from bl_operators.presets import PresetMenu
 
 from bpy.types import Panel
 
 
-class CYCLES_PT_sampling_presets(PresetPanel, Panel):
+class CYCLES_PT_sampling_presets(PresetMenu):
     bl_label = "Sampling Presets"
     preset_subdir = "cycles/sampling"
     preset_operator = "script.execute_preset"
@@ -31,7 +31,7 @@ class CYCLES_PT_sampling_presets(PresetPanel, Panel):
     COMPAT_ENGINES = {'CYCLES'}
 
 
-class CYCLES_PT_integrator_presets(PresetPanel, Panel):
+class CYCLES_PT_integrator_presets(PresetMenu):
     bl_label = "Integrator Presets"
     preset_subdir = "cycles/integrator"
     preset_operator = "script.execute_preset"
@@ -50,18 +50,14 @@ class CyclesButtonsPanel:
         return context.engine in cls.COMPAT_ENGINES
 
 
-# Adapt properties editor panel to display in node editor. We have to
-# copy the class rather than inherit due to the way bpy registration works.
-def node_panel(cls):
-    node_cls = type('NODE_' + cls.__name__, cls.__bases__, dict(cls.__dict__))
+class CyclesNodeButtonsPanel:
+    bl_space_type = "NODE_EDITOR"
+    bl_region_type = "UI"
+    COMPAT_ENGINES = {'CYCLES'}
 
-    node_cls.bl_space_type = 'NODE_EDITOR'
-    node_cls.bl_region_type = 'UI'
-    node_cls.bl_category = "Node"
-    if hasattr(node_cls, 'bl_parent_id'):
-        node_cls.bl_parent_id = 'NODE_' + node_cls.bl_parent_id
-
-    return node_cls
+    @classmethod
+    def poll(cls, context):
+        return context.engine in cls.COMPAT_ENGINES
 
 
 def get_device_type(context):
@@ -313,7 +309,7 @@ class CYCLES_RENDER_PT_subdivision(CyclesButtonsPanel, Panel):
 
     @classmethod
     def poll(self, context):
-        return (context.scene.render.engine == 'CYCLES') and (context.scene.cycles.feature_set == 'EXPERIMENTAL')
+        return context.scene.cycles.feature_set == 'EXPERIMENTAL'
 
     def draw(self, context):
         layout = self.layout
@@ -1119,7 +1115,7 @@ class CYCLES_PT_context_material(CyclesButtonsPanel, Panel):
             col.operator("object.material_slot_add", icon='ADD', text="")
             col.operator("object.material_slot_remove", icon='REMOVE', text="")
 
-            col.menu("MATERIAL_MT_context_menu", icon='DOWNARROW_HLT', text="")
+            col.menu("MATERIAL_MT_specials", icon='DOWNARROW_HLT', text="")
 
             if is_sortable:
                 col.separator()
@@ -1332,16 +1328,13 @@ class CYCLES_LIGHT_PT_light(CyclesButtonsPanel, Panel):
 
         light = context.light
         clamp = light.cycles
+        # cscene = context.scene.cycles
 
+        layout.prop(light, "type", expand=True)
+
+        layout.use_property_split = True
         layout.use_property_decorate = False
 
-        if self.bl_space_type == 'PROPERTIES':
-            layout.row().prop(light, "type", expand=True)
-            layout.use_property_split = True
-        else:
-            layout.use_property_split = True
-            layout.row().prop(light, "type")
-
         col = layout.column()
 
         if light.type in {'POINT', 'SUN', 'SPOT'}:
@@ -1620,8 +1613,7 @@ class CYCLES_MATERIAL_PT_preview(CyclesButtonsPanel, Panel):
 
     @classmethod
     def poll(cls, context):
-        mat = context.material
-        return mat and (not mat.grease_pencil) and CyclesButtonsPanel.poll(context)
+        return context.material and CyclesButtonsPanel.poll(context)
 
     def draw(self, context):
         self.layout.template_preview(context.material)
@@ -1633,8 +1625,7 @@ class CYCLES_MATERIAL_PT_surface(CyclesButtonsPanel, Panel):
 
     @classmethod
     def poll(cls, context):
-        mat = context.material
-        return mat and (not mat.grease_pencil) and CyclesButtonsPanel.poll(context)
+        return context.material and CyclesButtonsPanel.poll(context)
 
     def draw(self, context):
         layout = self.layout
@@ -1652,7 +1643,7 @@ class CYCLES_MATERIAL_PT_volume(CyclesButtonsPanel, Panel):
     @classmethod
     def poll(cls, context):
         mat = context.material
-        return mat and (not mat.grease_pencil) and mat.node_tree and CyclesButtonsPanel.poll(context)
+        return mat and mat.node_tree and CyclesButtonsPanel.poll(context)
 
     def draw(self, context):
         layout = self.layout
@@ -1670,7 +1661,7 @@ class CYCLES_MATERIAL_PT_displacement(CyclesButtonsPanel, Panel):
     @classmethod
     def poll(cls, context):
         mat = context.material
-        return mat and (not mat.grease_pencil) and mat.node_tree and CyclesButtonsPanel.poll(context)
+        return mat and mat.node_tree and CyclesButtonsPanel.poll(context)
 
     def draw(self, context):
         layout = self.layout
@@ -1686,8 +1677,7 @@ class CYCLES_MATERIAL_PT_settings(CyclesButtonsPanel, Panel):
 
     @classmethod
     def poll(cls, context):
-        mat = context.material
-        return mat and (not mat.grease_pencil) and CyclesButtonsPanel.poll(context)
+        return context.material and CyclesButtonsPanel.poll(context)
 
     @staticmethod
     def draw_shared(self, mat):
@@ -1939,6 +1929,7 @@ class CYCLES_RENDER_PT_debug(CyclesButtonsPanel, Panel):
         col = layout.column()
         col.label(text='OpenCL Flags:')
         col.prop(cscene, "debug_opencl_device_type", text="Device")
+        col.prop(cscene, "debug_opencl_kernel_single_program", text="Single Program")
         col.prop(cscene, "debug_use_opencl_debug", text="Debug")
         col.prop(cscene, "debug_opencl_mem_limit")
 
@@ -2042,6 +2033,43 @@ class CYCLES_RENDER_PT_simplify_culling(CyclesButtonsPanel, Panel):
         sub.prop(cscene, "distance_cull_margin", text="Distance")
 
 
+class CYCLES_NODE_PT_settings(CyclesNodeButtonsPanel, Panel):
+    bl_label = "Settings"
+    bl_category = "Node"
+    bl_options = {'DEFAULT_CLOSED'}
+
+    @classmethod
+    def poll(cls, context):
+        snode = context.space_data
+        return CyclesNodeButtonsPanel.poll(context) and \
+               snode.tree_type == 'ShaderNodeTree' and snode.id and \
+               snode.id.bl_rna.identifier == 'Material'
+
+    def draw(self, context):
+        material = context.space_data.id
+        CYCLES_MATERIAL_PT_settings.draw_shared(self, material)
+
+
+class CYCLES_NODE_PT_settings_surface(CyclesNodeButtonsPanel, Panel):
+    bl_label = "Surface"
+    bl_category = "Node"
+    bl_parent_id = "CYCLES_NODE_PT_settings"
+
+    def draw(self, context):
+        material = context.space_data.id
+        CYCLES_MATERIAL_PT_settings_surface.draw_shared(self, material)
+
+
+class CYCLES_NODE_PT_settings_volume(CyclesNodeButtonsPanel, Panel):
+    bl_label = "Volume"
+    bl_category = "Node"
+    bl_parent_id = "CYCLES_NODE_PT_settings"
+
+    def draw(self, context):
+        material = context.space_data.id
+        CYCLES_MATERIAL_PT_settings_volume.draw_shared(self, context, material)
+
+
 def draw_device(self, context):
     scene = context.scene
     layout = self.layout
@@ -2088,8 +2116,6 @@ def get_panels():
         'DATA_PT_spot',
         'MATERIAL_PT_context_material',
         'MATERIAL_PT_preview',
-        'NODE_DATA_PT_light',
-        'NODE_DATA_PT_spot',
         'VIEWLAYER_PT_filter',
         'VIEWLAYER_PT_layer_passes',
         'RENDER_PT_post_processing',
@@ -2175,15 +2201,9 @@ classes = (
     CYCLES_RENDER_PT_bake_selected_to_active,
     CYCLES_RENDER_PT_bake_output,
     CYCLES_RENDER_PT_debug,
-    node_panel(CYCLES_MATERIAL_PT_settings),
-    node_panel(CYCLES_MATERIAL_PT_settings_surface),
-    node_panel(CYCLES_MATERIAL_PT_settings_volume),
-    node_panel(CYCLES_WORLD_PT_ray_visibility),
-    node_panel(CYCLES_WORLD_PT_settings),
-    node_panel(CYCLES_WORLD_PT_settings_surface),
-    node_panel(CYCLES_WORLD_PT_settings_volume),
-    node_panel(CYCLES_LIGHT_PT_light),
-    node_panel(CYCLES_LIGHT_PT_spot),
+    CYCLES_NODE_PT_settings,
+    CYCLES_NODE_PT_settings_surface,
+    CYCLES_NODE_PT_settings_volume,
 )
 
 

intern/cycles/blender/addon/version_update.py

@@ -253,7 +253,7 @@ def do_versions(self):
                 pass
 
             # Init device list for UI
-            prop.get_devices(prop.compute_device_type)
+            prop.get_devices()
 
     # We don't modify startup file because it assumes to
     # have all the default values only.
@@ -267,7 +267,7 @@ def do_versions(self):
         library_versions.setdefault(library.version, []).append(library)
 
     # Do versioning per library, since they might have different versions.
-    max_need_versioning = (2, 80, 41)
+    max_need_versioning = (2, 79, 6)
     for version, libraries in library_versions.items():
         if version > max_need_versioning:
             continue

intern/cycles/blender/blender_curves.cpp

@@ -250,7 +250,7 @@ static bool ObtainCacheParticleUV(Mesh *mesh,
 					BL::Mesh::uv_layers_iterator l;
 					b_mesh->uv_layers.begin(l);
 
-					float2 uv = make_float2(0.0f, 0.0f);
+					float3 uv = make_float3(0.0f, 0.0f, 0.0f);
 					if(b_mesh->uv_layers.length())
 						b_psys.uv_on_emitter(psmd, *b_pa, pa_no, uv_num, &uv.x);
 					CData->curve_uv.push_back_slow(uv);
@@ -752,29 +752,45 @@ static void ExportCurveSegmentsMotion(Mesh *mesh, ParticleCurveData *CData, int
 static void ExportCurveTriangleUV(ParticleCurveData *CData,
                                   int vert_offset,
                                   int resol,
-                                  float2 *uvdata)
+                                  float3 *uvdata)
 {
 	if(uvdata == NULL)
 		return;
+
+	float time = 0.0f;
+	float prevtime = 0.0f;
+
 	int vertexindex = vert_offset;
 
 	for(int sys = 0; sys < CData->psys_firstcurve.size(); sys++) {
 		for(int curve = CData->psys_firstcurve[sys]; curve < CData->psys_firstcurve[sys] + CData->psys_curvenum[sys]; curve++) {
 			for(int curvekey = CData->curve_firstkey[curve]; curvekey < CData->curve_firstkey[curve] + CData->curve_keynum[curve] - 1; curvekey++) {
+				const float curve_time = CData->curvekey_time[curvekey];
+				const float curve_length = CData->curve_length[curve];
+				time = (curve_length > 0.0f) ? curve_time / curve_length : 0.0f;
+
 				for(int section = 0; section < resol; section++) {
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = prevtime;
 					vertexindex++;
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = time;
 					vertexindex++;
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = prevtime;
 					vertexindex++;
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = time;
 					vertexindex++;
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = prevtime;
 					vertexindex++;
 					uvdata[vertexindex] = CData->curve_uv[curve];
+					uvdata[vertexindex].z = time;
 					vertexindex++;
 				}
+
+				prevtime = time;
 			}
 		}
 	}
@@ -1047,9 +1063,9 @@ void BlenderSync::sync_curves(Mesh *mesh,
 					if(active_render)
 						attr_uv = mesh->attributes.add(std, name);
 					else
-						attr_uv = mesh->attributes.add(name, TypeFloat2, ATTR_ELEMENT_CORNER);
+						attr_uv = mesh->attributes.add(name, TypeDesc::TypePoint, ATTR_ELEMENT_CORNER);
 
-					float2 *uv = attr_uv->data_float2();
+					float3 *uv = attr_uv->data_float3();
 
 					ExportCurveTriangleUV(&CData, tri_num * 3, used_res, uv);
 				}
@@ -1057,9 +1073,9 @@ void BlenderSync::sync_curves(Mesh *mesh,
 					if(active_render)
 						attr_uv = mesh->curve_attributes.add(std, name);
 					else
-						attr_uv = mesh->curve_attributes.add(name, TypeFloat2,  ATTR_ELEMENT_CURVE);
+						attr_uv = mesh->curve_attributes.add(name, TypeDesc::TypePoint,  ATTR_ELEMENT_CURVE);
 
-					float2 *uv = attr_uv->data_float2();
+					float3 *uv = attr_uv->data_float3();
 
 					if(uv) {
 						size_t i = 0;

intern/cycles/blender/blender_mesh.cpp

@@ -66,7 +66,7 @@ struct MikkUserData {
 		else {
 			Attribute *attr_uv = attributes.find(ustring(layer_name));
 			if(attr_uv != NULL) {
-				texface = attr_uv->data_float2();
+				texface = attr_uv->data_float3();
 			}
 		}
 	}
@@ -75,7 +75,7 @@ struct MikkUserData {
 	int num_faces;
 
 	float3 *vertex_normal;
-	float2 *texface;
+	float3 *texface;
 	float3 *orco;
 	float3 orco_loc, orco_size;
 
@@ -150,7 +150,7 @@ static void mikk_get_texture_coordinate(const SMikkTSpaceContext *context,
 	const Mesh *mesh = userdata->mesh;
 	if(userdata->texface != NULL) {
 		const int corner_index = mikk_corner_index(mesh, face_num, vert_num);
-		float2 tfuv = userdata->texface[corner_index];
+		float3 tfuv = userdata->texface[corner_index];
 		uv[0] = tfuv.x;
 		uv[1] = tfuv.y;
 	}
@@ -430,18 +430,18 @@ static void attr_create_uv_map(Scene *scene,
 				}
 				else {
 					uv_attr = mesh->attributes.add(uv_name,
-					                               TypeFloat2,
+					                               TypeDesc::TypePoint,
 					                               ATTR_ELEMENT_CORNER);
 				}
 
 				BL::Mesh::loop_triangles_iterator t;
-				float2 *fdata = uv_attr->data_float2();
+				float3 *fdata = uv_attr->data_float3();
 
 				for(b_mesh.loop_triangles.begin(t); t != b_mesh.loop_triangles.end(); ++t) {
 					int3 li = get_int3(t->loops());
-					fdata[0] = get_float2(l->data[li[0]].uv());
-					fdata[1] = get_float2(l->data[li[1]].uv());
-					fdata[2] = get_float2(l->data[li[2]].uv());
+					fdata[0] = get_float3(l->data[li[0]].uv());
+					fdata[1] = get_float3(l->data[li[1]].uv());
+					fdata[2] = get_float3(l->data[li[2]].uv());
 					fdata += 3;
 				}
 			}
@@ -509,19 +509,19 @@ static void attr_create_subd_uv_map(Scene *scene,
 				if(active_render)
 					uv_attr = mesh->subd_attributes.add(uv_std, uv_name);
 				else
-					uv_attr = mesh->subd_attributes.add(uv_name, TypeFloat2, ATTR_ELEMENT_CORNER);
+					uv_attr = mesh->subd_attributes.add(uv_name, TypeDesc::TypePoint, ATTR_ELEMENT_CORNER);
 
 				if(subdivide_uvs) {
 					uv_attr->flags |= ATTR_SUBDIVIDED;
 				}
 
 				BL::Mesh::polygons_iterator p;
-				float2 *fdata = uv_attr->data_float2();
+				float3 *fdata = uv_attr->data_float3();
 
 				for(b_mesh.polygons.begin(p); p != b_mesh.polygons.end(); ++p) {
 					int n = p->loop_total();
 					for(int j = 0; j < n; j++) {
-						*(fdata++) = get_float2(l->data[p->loop_start() + j].uv());
+						*(fdata++) = get_float3(l->data[p->loop_start() + j].uv());
 					}
 				}
 			}
@@ -927,6 +927,8 @@ static void create_subd_mesh(Scene *scene,
 	sdparams.dicing_rate = max(0.1f, RNA_float_get(&cobj, "dicing_rate") * dicing_rate);
 	sdparams.max_level = max_subdivisions;
 
+	scene->dicing_camera->update(scene);
+	sdparams.camera = scene->dicing_camera;
 	sdparams.objecttoworld = get_transform(b_ob.matrix_world());
 }
 
@@ -1037,8 +1039,6 @@ Mesh *BlenderSync::sync_mesh(BL::Depsgraph& b_depsgraph,
 	if(mesh_synced.find(mesh) != mesh_synced.end())
 		return mesh;
 
-	progress.set_sync_status("Synchronizing object", b_ob.name());
-
 	mesh_synced.insert(mesh);
 
 	/* create derived mesh */

intern/cycles/blender/blender_object.cpp

@@ -349,7 +349,7 @@ Object *BlenderSync::sync_object(BL::Depsgraph& b_depsgraph,
 	/* Visibility flags for both parent and child. */
 	PointerRNA cobject = RNA_pointer_get(&b_ob.ptr, "cycles");
 	bool use_holdout = get_boolean(cobject, "is_holdout") ||
-	                   b_parent.holdout_get(PointerRNA_NULL, b_view_layer);
+	                   b_parent.holdout_get(b_view_layer);
 	uint visibility = object_ray_visibility(b_ob) & PATH_RAY_ALL_VISIBILITY;
 
 	if(b_parent.ptr.data != b_ob.ptr.data) {
@@ -364,7 +364,7 @@ Object *BlenderSync::sync_object(BL::Depsgraph& b_depsgraph,
 #endif
 
 	/* Clear camera visibility for indirect only objects. */
-	bool use_indirect_only = b_parent.indirect_only_get(PointerRNA_NULL, b_view_layer);
+	bool use_indirect_only = b_parent.indirect_only_get(b_view_layer);
 	if(use_indirect_only) {
 		visibility &= ~PATH_RAY_CAMERA;
 	}
@@ -544,6 +544,8 @@ void BlenderSync::sync_objects(BL::Depsgraph& b_depsgraph, float motion_time)
 		BL::DepsgraphObjectInstance b_instance = *b_instance_iter;
 		BL::Object b_ob = b_instance.object();
 
+		progress.set_sync_status("Synchronizing object", b_ob.name());
+
 		/* load per-object culling data */
 		culling.init_object(scene, b_ob);
 

intern/cycles/blender/blender_python.cpp

@@ -23,7 +23,6 @@
 #include "blender/blender_session.h"
 
 #include "render/denoising.h"
-#include "render/merge.h"
 
 #include "util/util_debug.h"
 #include "util/util_foreach.h"
@@ -105,6 +104,7 @@ bool debug_flags_sync_from_scene(BL::Scene b_scene)
 	/* Synchronize other OpenCL flags. */
 	flags.opencl.debug = get_boolean(cscene, "debug_use_opencl_debug");
 	flags.opencl.mem_limit = ((size_t)get_int(cscene, "debug_opencl_mem_limit"))*1024*1024;
+	flags.opencl.single_program = get_boolean(cscene, "debug_opencl_kernel_single_program");
 	return flags.opencl.device_type != opencl_device_type;
 }
 
@@ -643,8 +643,9 @@ static PyObject *opencl_compile_func(PyObject * /*self*/, PyObject *args)
 }
 #endif
 
-static bool image_parse_filepaths(PyObject *pyfilepaths, vector<string>& filepaths)
+static bool denoise_parse_filepaths(PyObject *pyfilepaths, vector<string>& filepaths)
 {
+
 	if(PyUnicode_Check(pyfilepaths)) {
 		const char *filepath = PyUnicode_AsUTF8(pyfilepaths);
 		filepaths.push_back(filepath);
@@ -713,12 +714,12 @@ static PyObject *denoise_func(PyObject * /*self*/, PyObject *args, PyObject *key
 	/* Parse file paths list. */
 	vector<string> input, output;
 
-	if(!image_parse_filepaths(pyinput, input)) {
+	if(!denoise_parse_filepaths(pyinput, input)) {
 		return NULL;
 	}
 
 	if(pyoutput) {
-		if(!image_parse_filepaths(pyoutput, output)) {
+		if(!denoise_parse_filepaths(pyoutput, output)) {
 			return NULL;
 		}
 	}
@@ -757,42 +758,6 @@ static PyObject *denoise_func(PyObject * /*self*/, PyObject *args, PyObject *key
 	Py_RETURN_NONE;
 }
 
-static PyObject *merge_func(PyObject * /*self*/, PyObject *args, PyObject *keywords)
-{
-	static const char *keyword_list[] = {"input", "output", NULL};
-	PyObject *pyinput, *pyoutput = NULL;
-
-	if (!PyArg_ParseTupleAndKeywords(args, keywords, "OO", (char**)keyword_list, &pyinput, &pyoutput)) {
-		return NULL;
-	}
-
-	/* Parse input list. */
-	vector<string> input;
-	if(!image_parse_filepaths(pyinput, input)) {
-		return NULL;
-	}
-
-	/* Parse output string. */
-	if(!PyUnicode_Check(pyoutput)) {
-		PyErr_SetString(PyExc_ValueError, "Output must be a string.");
-		return NULL;
-	}
-	string output = PyUnicode_AsUTF8(pyoutput);
-
-	/* Merge. */
-	ImageMerger merger;
-	merger.input = input;
-	merger.output = output;
-
-	if(!merger.run()) {
-		PyErr_SetString(PyExc_ValueError, merger.error.c_str());
-		return NULL;
-	}
-
-	Py_RETURN_NONE;
-}
-
-
 static PyObject *debug_flags_update_func(PyObject * /*self*/, PyObject *args)
 {
 	PyObject *pyscene;
@@ -956,7 +921,6 @@ static PyMethodDef methods[] = {
 
 	/* Standalone denoising */
 	{"denoise", (PyCFunction)denoise_func, METH_VARARGS|METH_KEYWORDS, ""},
-	{"merge", (PyCFunction)merge_func, METH_VARARGS|METH_KEYWORDS, ""},
 
 	/* Debugging routines */
 	{"debug_flags_update", debug_flags_update_func, METH_VARARGS, ""},

intern/cycles/blender/blender_session.cpp

@@ -416,26 +416,16 @@ void BlenderSession::stamp_view_layer_metadata(Scene *scene, const string& view_
 	/* Write cryptomatte metadata. */
 	if(scene->film->cryptomatte_passes & CRYPT_OBJECT) {
 		add_cryptomatte_layer(b_rr, view_layer_name + ".CryptoObject",
-		                      scene->object_manager->get_cryptomatte_objects(scene));
+							  scene->object_manager->get_cryptomatte_objects(scene));
 	}
 	if(scene->film->cryptomatte_passes & CRYPT_MATERIAL) {
 		add_cryptomatte_layer(b_rr, view_layer_name + ".CryptoMaterial",
-		                      scene->shader_manager->get_cryptomatte_materials(scene));
+							  scene->shader_manager->get_cryptomatte_materials(scene));
 	}
 	if(scene->film->cryptomatte_passes & CRYPT_ASSET) {
 		add_cryptomatte_layer(b_rr, view_layer_name + ".CryptoAsset",
-		                      scene->object_manager->get_cryptomatte_assets(scene));
+							  scene->object_manager->get_cryptomatte_assets(scene));
 	}
-
-	/* Store synchronization and bare-render times. */
-	double total_time, render_time;
-	session->progress.get_time(total_time, render_time);
-	b_rr.stamp_data_add_field((prefix + "total_time").c_str(),
-	                          time_human_readable_from_seconds(total_time).c_str());
-	b_rr.stamp_data_add_field((prefix + "render_time").c_str(),
-	                          time_human_readable_from_seconds(render_time).c_str());
-	b_rr.stamp_data_add_field((prefix + "synchronization_time").c_str(),
-	                          time_human_readable_from_seconds(total_time - render_time).c_str());
 }
 
 void BlenderSession::render(BL::Depsgraph& b_depsgraph_)
@@ -461,7 +451,7 @@ void BlenderSession::render(BL::Depsgraph& b_depsgraph_)
 	b_rlay_name = b_view_layer.name();
 
 	/* add passes */
-	vector<Pass> passes = sync->sync_render_passes(b_rlay, b_view_layer);
+	vector<Pass> passes = sync->sync_render_passes(b_rlay, b_view_layer, session_params);
 	buffer_params.passes = passes;
 
 	PointerRNA crl = RNA_pointer_get(&b_view_layer.ptr, "cycles");
@@ -978,11 +968,6 @@ void BlenderSession::get_status(string& status, string& substatus)
 	session->progress.get_status(status, substatus);
 }
 
-void BlenderSession::get_kernel_status(string& kernel_status)
-{
-	session->progress.get_kernel_status(kernel_status);
-}
-
 void BlenderSession::get_progress(float& progress, double& total_time, double& render_time)
 {
 	session->progress.get_time(total_time, render_time);
@@ -1001,15 +986,15 @@ void BlenderSession::update_bake_progress()
 
 void BlenderSession::update_status_progress()
 {
-	string timestatus, status, substatus, kernel_status;
+	string timestatus, status, substatus;
 	string scene_status = "";
 	float progress;
 	double total_time, remaining_time = 0, render_time;
+	char time_str[128];
 	float mem_used = (float)session->stats.mem_used / 1024.0f / 1024.0f;
 	float mem_peak = (float)session->stats.mem_peak / 1024.0f / 1024.0f;
 
 	get_status(status, substatus);
-	get_kernel_status(kernel_status);
 	get_progress(progress, total_time, render_time);
 
 	if(progress > 0)
@@ -1024,7 +1009,8 @@ void BlenderSession::update_status_progress()
 			scene_status += ", " + b_rview_name;
 
 		if(remaining_time > 0) {
-			timestatus += "Remaining:" + time_human_readable_from_seconds(remaining_time) + " | ";
+			BLI_timecode_string_from_time_simple(time_str, sizeof(time_str), remaining_time);
+			timestatus += "Remaining:" + string(time_str) + " | ";
 		}
 
 		timestatus += string_printf("Mem:%.2fM, Peak:%.2fM", (double)mem_used, (double)mem_peak);
@@ -1033,8 +1019,6 @@ void BlenderSession::update_status_progress()
 			status = " | " + status;
 		if(substatus.size() > 0)
 			status += " | " + substatus;
-		if(kernel_status.size() > 0)
-			status += " | " + kernel_status;
 	}
 
 	double current_time = time_dt();

intern/cycles/blender/blender_session.h

@@ -90,7 +90,6 @@ public:
 	void tag_redraw();
 	void tag_update();
 	void get_status(string& status, string& substatus);
-	void get_kernel_status(string& kernel_status);
 	void get_progress(float& progress, double& total_time, double& render_time);
 	void test_cancel();
 	void update_status_progress();

intern/cycles/blender/blender_shader.cpp

@@ -1050,18 +1050,13 @@ static void add_nodes(Scene *scene,
 				graph->add(proxy);
 			}
 		}
-		else if(b_node->is_a(&RNA_ShaderNodeGroup) ||
-		        b_node->is_a(&RNA_NodeCustomGroup) ||
-		        b_node->is_a(&RNA_ShaderNodeCustomGroup)) {
+		else if(b_node->is_a(&RNA_ShaderNodeGroup) || b_node->is_a(&RNA_NodeCustomGroup)) {
 
 			BL::ShaderNodeTree b_group_ntree(PointerRNA_NULL);
 			if(b_node->is_a(&RNA_ShaderNodeGroup))
 				b_group_ntree = BL::ShaderNodeTree(((BL::NodeGroup)(*b_node)).node_tree());
-			else if (b_node->is_a(&RNA_NodeCustomGroup))
-				b_group_ntree = BL::ShaderNodeTree(((BL::NodeCustomGroup)(*b_node)).node_tree());
 			else
-				b_group_ntree = BL::ShaderNodeTree(((BL::ShaderNodeCustomGroup)(*b_node)).node_tree());
-
+				b_group_ntree = BL::ShaderNodeTree(((BL::NodeCustomGroup)(*b_node)).node_tree());
 			ProxyMap group_proxy_input_map, group_proxy_output_map;
 
 			/* Add a proxy node for each socket

intern/cycles/blender/blender_sync.cpp

@@ -85,11 +85,10 @@ void BlenderSync::sync_recalc(BL::Depsgraph& b_depsgraph)
 	 * so we can do it later on if doing it immediate is not suitable. */
 
 	bool has_updated_objects = b_depsgraph.id_type_updated(BL::DriverTarget::id_type_OBJECT);
+	bool dicing_prop_changed = false;
 
 	if(experimental) {
-		/* Mark all meshes as needing to be exported again if dicing changed. */
 		PointerRNA cscene = RNA_pointer_get(&b_scene.ptr, "cycles");
-		bool dicing_prop_changed = false;
 
 		float updated_dicing_rate = preview ? RNA_float_get(&cscene, "preview_dicing_rate")
 		                                    : RNA_float_get(&cscene, "dicing_rate");
@@ -105,15 +104,6 @@ void BlenderSync::sync_recalc(BL::Depsgraph& b_depsgraph)
 			max_subdivisions = updated_max_subdivisions;
 			dicing_prop_changed = true;
 		}
-
-		if(dicing_prop_changed) {
-			for(const pair<void*, Mesh*>& iter: mesh_map.key_to_scene_data()) {
-				Mesh *mesh = iter.second;
-				if(mesh->subdivision_type != Mesh::SUBDIVISION_NONE) {
-					mesh_map.set_recalc(iter.first);
-				}
-			}
-		}
 	}
 
 	/* Iterate over all IDs in this depsgraph. */
@@ -143,7 +133,7 @@ void BlenderSync::sync_recalc(BL::Depsgraph& b_depsgraph)
 
 			if(object_is_mesh(b_ob)) {
 				if(updated_geometry ||
-				   (object_subdivision_type(b_ob, preview, experimental) != Mesh::SUBDIVISION_NONE))
+				   (dicing_prop_changed && object_subdivision_type(b_ob, preview, experimental) != Mesh::SUBDIVISION_NONE))
 				{
 					BL::ID key = BKE_object_is_modified(b_ob)? b_ob: b_ob.data();
 					mesh_map.set_recalc(key);
@@ -514,11 +504,16 @@ int BlenderSync::get_denoising_pass(BL::RenderPass& b_pass)
 }
 
 vector<Pass> BlenderSync::sync_render_passes(BL::RenderLayer& b_rlay,
-                                             BL::ViewLayer& b_view_layer)
+                                             BL::ViewLayer& b_view_layer,
+                                             const SessionParams &session_params)
 {
 	vector<Pass> passes;
 	Pass::add(PASS_COMBINED, passes);
 
+	if(!session_params.device.advanced_shading) {
+		return passes;
+	}
+
 	/* loop over passes */
 	BL::RenderLayer::passes_iterator b_pass_iter;
 
@@ -774,9 +769,6 @@ SessionParams BlenderSync::get_session_params(BL::RenderEngine& b_engine,
 		}
 	}
 
-	/* Clamp samples. */
-	params.samples = min(params.samples, Integrator::MAX_SAMPLES);
-
 	/* tiles */
 	const bool is_cpu = (params.device.type == DEVICE_CPU);
 	if(!is_cpu && !background) {

intern/cycles/blender/blender_sync.h

@@ -68,7 +68,8 @@ public:
 	               void **python_thread_state);
 	void sync_view_layer(BL::SpaceView3D& b_v3d, BL::ViewLayer& b_view_layer);
 	vector<Pass> sync_render_passes(BL::RenderLayer& b_render_layer,
-	                                BL::ViewLayer& b_view_layer);
+	                                BL::ViewLayer& b_view_layer,
+	                                const SessionParams &session_params);
 	void sync_integrator();
 	void sync_camera(BL::RenderSettings& b_render,
 	                 BL::Object& b_override,

intern/cycles/blender/blender_util.h

@@ -32,6 +32,7 @@
  * todo: clean this up ... */
 
 extern "C" {
+size_t BLI_timecode_string_from_time_simple(char *str, size_t maxlen, double time_seconds);
 void BKE_image_user_frame_calc(void *iuser, int cfra);
 void BKE_image_user_file_path(void *iuser, void *ima, char *path);
 unsigned char *BKE_image_get_pixels_for_frame(void *image, int frame);
@@ -323,6 +324,46 @@ static inline int4 get_int4(const BL::Array<int, 4>& array)
 	return make_int4(array[0], array[1], array[2], array[3]);
 }
 
+static inline uint get_layer(const BL::Array<bool, 20>& array)
+{
+	uint layer = 0;
+
+	for(uint i = 0; i < 20; i++)
+		if(array[i])
+			layer |= (1 << i);
+
+	return layer;
+}
+
+static inline uint get_layer(const BL::Array<bool, 20>& array,
+                             const BL::Array<bool, 8>& local_array,
+                             bool is_light = false,
+                             uint view_layers = (1 << 20) - 1)
+{
+	uint layer = 0;
+
+	for(uint i = 0; i < 20; i++)
+		if(array[i])
+			layer |= (1 << i);
+
+	if(is_light) {
+		/* Consider light is visible if it was visible without layer
+		 * override, which matches behavior of Blender Internal.
+		 */
+		if(layer & view_layers) {
+			for(uint i = 0; i < 8; i++)
+				layer |= (1 << (20+i));
+		}
+	}
+	else {
+		for(uint i = 0; i < 8; i++)
+			if(local_array[i])
+				layer |= (1 << (20+i));
+	}
+
+	return layer;
+}
+
 static inline float3 get_float3(PointerRNA& ptr, const char *name)
 {
 	float3 f;
@@ -628,11 +669,6 @@ public:
 		b_recalc.insert(id.ptr.data);
 	}
 
-	void set_recalc(void *id_ptr)
-	{
-		b_recalc.insert(id_ptr);
-	}
-
 	bool has_recalc()
 	{
 		return !(b_recalc.empty());
@@ -728,11 +764,6 @@ public:
 		return deleted;
 	}
 
-	const map<K, T*>& key_to_scene_data()
-	{
-		return b_map;
-	}
-
 protected:
 	vector<T*> *scene_data;
 	map<K, T*> b_map;

intern/cycles/bvh/bvh_embree.cpp

@@ -149,13 +149,6 @@ static void rtc_filter_occluded_func(const RTCFilterFunctionNArguments* args)
 				break;
 			}
 
-			/* Ignore curves. */
-			if(hit->geomID & 1) {
-				/* This tells Embree to continue tracing. */
-				*args->valid = 0;
-				break;
-			}
-
 			/* See triangle_intersect_subsurface() for the native equivalent. */
 			for(int i = min(ctx->max_hits, ctx->ss_isect->num_hits) - 1; i >= 0; --i) {
 				if(ctx->ss_isect->hits[i].t == ray->tfar) {
@@ -396,45 +389,6 @@ void BVHEmbree::build(Progress& progress, Stats *stats_)
 	               (params.use_spatial_split ? RTC_BUILD_QUALITY_HIGH : RTC_BUILD_QUALITY_MEDIUM);
 	rtcSetSceneBuildQuality(scene, build_quality);
 
-	/* Count triangles and curves first, reserve arrays once. */
-	size_t prim_count = 0;
-
-	foreach(Object *ob, objects) {
-		if (params.top_level) {
-			if (!ob->is_traceable()) {
-				continue;
-			}
-			if (!ob->mesh->is_instanced()) {
-				if(params.primitive_mask & PRIMITIVE_ALL_TRIANGLE) {
-					prim_count += ob->mesh->num_triangles();
-				}
-				if (params.primitive_mask & PRIMITIVE_ALL_CURVE) {
-					for (size_t j = 0; j < ob->mesh->num_curves(); ++j) {
-						prim_count += ob->mesh->get_curve(j).num_segments();
-					}
-				}
-			}
-			else {
-				++prim_count;
-			}
-		}
-		else {
-			if (params.primitive_mask & PRIMITIVE_ALL_TRIANGLE && ob->mesh->num_triangles() > 0) {
-				prim_count += ob->mesh->num_triangles();
-			}
-			if (params.primitive_mask & PRIMITIVE_ALL_CURVE) {
-				for (size_t j = 0; j < ob->mesh->num_curves(); ++j) {
-					prim_count += ob->mesh->get_curve(j).num_segments();
-				}
-			}
-		}
-	}
-
-	pack.prim_object.reserve(prim_count);
-	pack.prim_type.reserve(prim_count);
-	pack.prim_index.reserve(prim_count);
-	pack.prim_tri_index.reserve(prim_count);
-
 	int i = 0;
 
 	pack.object_node.clear();
@@ -576,20 +530,15 @@ void BVHEmbree::add_triangles(Object *ob, int i)
 
 	update_tri_vertex_buffer(geom_id, mesh);
 
-	size_t prim_object_size = pack.prim_object.size();
-	pack.prim_object.resize(prim_object_size + num_triangles);
-	size_t prim_type_size = pack.prim_type.size();
-	pack.prim_type.resize(prim_type_size + num_triangles);
-	size_t prim_index_size = pack.prim_index.size();
-	pack.prim_index.resize(prim_index_size + num_triangles);
-	pack.prim_tri_index.resize(prim_index_size + num_triangles);
-	int prim_type = (num_motion_steps > 1 ? PRIMITIVE_MOTION_TRIANGLE : PRIMITIVE_TRIANGLE);
-
+	pack.prim_object.reserve(pack.prim_object.size() + num_triangles);
+	pack.prim_type.reserve(pack.prim_type.size() + num_triangles);
+	pack.prim_index.reserve(pack.prim_index.size() + num_triangles);
+	pack.prim_tri_index.reserve(pack.prim_index.size() + num_triangles);
 	for(size_t j = 0; j < num_triangles; ++j) {
-		pack.prim_object[prim_object_size + j] = i;
-		pack.prim_type[prim_type_size + j] = prim_type;
-		pack.prim_index[prim_index_size + j] = j;
-		pack.prim_tri_index[prim_index_size + j] = j;
+		pack.prim_object.push_back_reserved(i);
+		pack.prim_type.push_back_reserved(num_motion_steps > 1 ? PRIMITIVE_MOTION_TRIANGLE : PRIMITIVE_TRIANGLE);
+		pack.prim_index.push_back_reserved(j);
+		pack.prim_tri_index.push_back_reserved(j);
 	}
 
 	rtcSetGeometryUserData(geom_id, (void*) prim_offset);
@@ -680,7 +629,7 @@ void BVHEmbree::update_curve_vertex_buffer(RTCGeometry geom_id, const Mesh* mesh
 		float4 *rtc_tangents = NULL;
 		if(use_curves) {
 			rtc_tangents = (float4*)rtcSetNewGeometryBuffer(geom_id, RTC_BUFFER_TYPE_TANGENT, t,
-			                                                RTC_FORMAT_FLOAT4, sizeof (float) * 4, num_keys);
+																RTC_FORMAT_FLOAT4, sizeof (float) * 4, num_keys);
 			assert(rtc_tangents);
 		}
 		assert(rtc_verts);
@@ -742,14 +691,10 @@ void BVHEmbree::add_curves(Object *ob, int i)
 	}
 
 	/* Make room for Cycles specific data. */
-	size_t prim_object_size = pack.prim_object.size();
-	pack.prim_object.resize(prim_object_size + num_segments);
-	size_t prim_type_size = pack.prim_type.size();
-	pack.prim_type.resize(prim_type_size + num_segments);
-	size_t prim_index_size = pack.prim_index.size();
-	pack.prim_index.resize(prim_index_size + num_segments);
-	size_t prim_tri_index_size = pack.prim_index.size();
-	pack.prim_tri_index.resize(prim_tri_index_size + num_segments);
+	pack.prim_object.reserve(pack.prim_object.size() + num_segments);
+	pack.prim_type.reserve(pack.prim_type.size() + num_segments);
+	pack.prim_index.reserve(pack.prim_index.size() + num_segments);
+	pack.prim_tri_index.reserve(pack.prim_index.size() + num_segments);
 
 	enum RTCGeometryType type = (!use_curves) ? RTC_GEOMETRY_TYPE_FLAT_LINEAR_CURVE :
 	                            (use_ribbons ? RTC_GEOMETRY_TYPE_FLAT_HERMITE_CURVE :
@@ -758,18 +703,18 @@ void BVHEmbree::add_curves(Object *ob, int i)
 	RTCGeometry geom_id = rtcNewGeometry(rtc_shared_device, type);
 	rtcSetGeometryTessellationRate(geom_id, curve_subdivisions);
 	unsigned *rtc_indices = (unsigned*) rtcSetNewGeometryBuffer(geom_id, RTC_BUFFER_TYPE_INDEX, 0,
-	                                                            RTC_FORMAT_UINT, sizeof (int), num_segments);
+																RTC_FORMAT_UINT, sizeof (int), num_segments);
 	size_t rtc_index = 0;
 	for(size_t j = 0; j < num_curves; ++j) {
 		Mesh::Curve c = mesh->get_curve(j);
 		for(size_t k = 0; k < c.num_segments(); ++k) {
 			rtc_indices[rtc_index] = c.first_key + k;
 			/* Cycles specific data. */
-			pack.prim_object[prim_object_size + rtc_index] = i;
-			pack.prim_type[prim_type_size + rtc_index] = (PRIMITIVE_PACK_SEGMENT(num_motion_steps > 1 ?
-			                                              PRIMITIVE_MOTION_CURVE : PRIMITIVE_CURVE, k));
-			pack.prim_index[prim_index_size + rtc_index] = j;
-			pack.prim_tri_index[prim_tri_index_size + rtc_index] = rtc_index;
+			pack.prim_object.push_back_reserved(i);
+			pack.prim_type.push_back_reserved(PRIMITIVE_PACK_SEGMENT(num_motion_steps > 1 ?
+																	 PRIMITIVE_MOTION_CURVE : PRIMITIVE_CURVE, k));
+			pack.prim_index.push_back_reserved(j);
+			pack.prim_tri_index.push_back_reserved(rtc_index);
 
 			++rtc_index;
 		}

intern/cycles/device/device.cpp

@@ -349,7 +349,6 @@ void Device::draw_pixels(
 		draw_params.unbind_display_space_shader_cb();
 	}
 
-	glDeleteVertexArrays(1, &vertex_array_object);
 	glBindTexture(GL_TEXTURE_2D, 0);
 	glDeleteTextures(1, &texid);
 

intern/cycles/device/device.h

@@ -56,14 +56,6 @@ enum DeviceTypeMask {
 	DEVICE_MASK_ALL = ~0
 };
 
-enum DeviceKernelStatus {
-	DEVICE_KERNEL_WAITING_FOR_FEATURE_KERNEL = 0,
-	DEVICE_KERNEL_FEATURE_KERNEL_AVAILABLE,
-	DEVICE_KERNEL_USING_FEATURE_KERNEL,
-	DEVICE_KERNEL_FEATURE_KERNEL_INVALID,
-	DEVICE_KERNEL_UNKNOWN,
-};
-
 #define DEVICE_MASK(type) (DeviceTypeMask)(1 << type)
 
 class DeviceInfo {
@@ -73,6 +65,7 @@ public:
 	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 advanced_shading;          /* Supports full shading system. */
 	bool has_half_images;           /* Support half-float textures. */
 	bool has_volume_decoupled;      /* Decoupled volume shading. */
 	bool has_osl;                   /* Support Open Shading Language. */
@@ -88,6 +81,7 @@ public:
 		num = 0;
 		cpu_threads = 0;
 		display_device = false;
+		advanced_shading = true;
 		has_half_images = false;
 		has_volume_decoupled = false;
 		has_osl = false;
@@ -340,20 +334,6 @@ public:
 	        const DeviceRequestedFeatures& /*requested_features*/)
 	{ return true; }
 
-	/* Wait for device to become available to upload data and receive tasks
-	 * This method is used by the OpenCL device to load the
-	 * optimized kernels or when not (yet) available load the
-	 * generic kernels (only during foreground rendering) */
-	virtual bool wait_for_availability(
-	        const DeviceRequestedFeatures& /*requested_features*/)
-	{ return true; }
-	/* Check if there are 'better' kernels available to be used
-	 * We can switch over to these kernels
-	 * This method is used to determine if we can switch the preview kernels
-	 * to regular kernels */
-	virtual DeviceKernelStatus get_active_kernel_switch_state()
-	{ return DEVICE_KERNEL_USING_FEATURE_KERNEL; }
-
 	/* tasks */
 	virtual int get_split_task_count(DeviceTask& task) = 0;
 	virtual void task_add(DeviceTask& task) = 0;

intern/cycles/device/device_cpu.cpp

@@ -761,8 +761,8 @@ public:
 		int start_sample = tile.start_sample;
 		int end_sample = tile.start_sample + tile.num_samples;
 
-		/* Needed for Embree. */
-		SIMD_SET_FLUSH_TO_ZERO;
+		_MM_SET_FLUSH_ZERO_MODE(_MM_FLUSH_ZERO_ON);
+		_MM_SET_DENORMALS_ZERO_MODE(_MM_DENORMALS_ZERO_ON);
 
 		for(int sample = start_sample; sample < end_sample; sample++) {
 			if(task.get_cancel() || task_pool.canceled()) {
@@ -1123,6 +1123,7 @@ void device_cpu_info(vector<DeviceInfo>& devices)
 	info.description = system_cpu_brand_string();
 	info.id = "CPU";
 	info.num = 0;
+	info.advanced_shading = true;
 	info.has_volume_decoupled = true;
 	info.has_osl = true;
 	info.has_half_images = true;

intern/cycles/device/device_cuda.cpp

@@ -400,9 +400,9 @@ public:
 			       major, minor);
 			return false;
 		}
-		else if(cuda_version != 101) {
+		else if(cuda_version != 80) {
 			printf("CUDA version %d.%d detected, build may succeed but only "
-			       "CUDA 10.1 is officially supported.\n",
+			       "CUDA 8.0 is officially supported.\n",
 			       major, minor);
 		}
 		return true;
@@ -2532,6 +2532,7 @@ void device_cuda_info(vector<DeviceInfo>& devices)
 		info.description = string(name);
 		info.num = num;
 
+		info.advanced_shading = (major >= 3);
 		info.has_half_images = (major >= 3);
 		info.has_volume_decoupled = false;
 

intern/cycles/device/device_multi.cpp

@@ -120,38 +120,6 @@ public:
 		return true;
 	}
 
-	bool wait_for_availability(const DeviceRequestedFeatures& requested_features)
-	{
-		foreach(SubDevice& sub, devices)
-			if(!sub.device->wait_for_availability(requested_features))
-				return false;
-
-		return true;
-	}
-
-	DeviceKernelStatus get_active_kernel_switch_state()
-	{
-		DeviceKernelStatus result = DEVICE_KERNEL_USING_FEATURE_KERNEL;
-
-		foreach(SubDevice& sub, devices) {
-			DeviceKernelStatus subresult = sub.device->get_active_kernel_switch_state();
-			switch (subresult) {
-				case DEVICE_KERNEL_WAITING_FOR_FEATURE_KERNEL:
-					result = subresult;
-					break;
-
-				case DEVICE_KERNEL_FEATURE_KERNEL_INVALID:
-				case DEVICE_KERNEL_FEATURE_KERNEL_AVAILABLE:
-					return subresult;
-
-				case DEVICE_KERNEL_USING_FEATURE_KERNEL:
-				case DEVICE_KERNEL_UNKNOWN:
-					break;
-			}
-		}
-		return result;
-	}
-
 	void mem_alloc(device_memory& mem)
 	{
 		device_ptr key = unique_key++;

intern/cycles/device/device_network.cpp

@@ -308,6 +308,7 @@ void device_network_info(vector<DeviceInfo>& devices)
 	info.num = 0;
 
 	/* todo: get this info from device */
+	info.advanced_shading = true;
 	info.has_volume_decoupled = false;
 	info.has_osl = false;
 

intern/cycles/device/device_opencl.cpp

@@ -119,6 +119,7 @@ void device_opencl_info(vector<DeviceInfo>& devices)
 		info.num = num_devices;
 		/* We don't know if it's used for display, but assume it is. */
 		info.display_device = true;
+		info.advanced_shading = OpenCLInfo::kernel_use_advanced_shading(platform_name);
 		info.use_split_kernel = true;
 		info.has_volume_decoupled = false;
 		info.id = id;

intern/cycles/device/opencl/opencl.h

@@ -84,6 +84,7 @@ class OpenCLInfo
 public:
 	static cl_device_type device_type();
 	static bool use_debug();
+	static bool kernel_use_advanced_shading(const string& platform_name);
 	static bool device_supported(const string& platform_name,
 	                             const cl_device_id device_id);
 	static bool platform_version_check(cl_platform_id platform,
@@ -94,6 +95,7 @@ public:
 	                              cl_device_id device_id);
 	static void get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices,
 	                               bool force_all = false);
+	static bool use_single_program();
 
 	/* ** Some handy shortcuts to low level cl*GetInfo() functions. ** */
 
@@ -260,22 +262,16 @@ class OpenCLDevice : public Device
 {
 public:
 	DedicatedTaskPool task_pool;
-
-	/* Task pool for required kernels (base, AO kernels during foreground rendering) */
-	TaskPool load_required_kernel_task_pool;
-	/* Task pool for optional kernels (feature kernels during foreground rendering) */
-	TaskPool load_kernel_task_pool;
 	cl_context cxContext;
 	cl_command_queue cqCommandQueue;
 	cl_platform_id cpPlatform;
 	cl_device_id cdDevice;
 	cl_int ciErr;
 	int device_num;
-	bool use_preview_kernels;
 
 	class OpenCLProgram {
 	public:
-		OpenCLProgram() : loaded(false), needs_compiling(true), program(NULL), device(NULL) {}
+		OpenCLProgram() : loaded(false), program(NULL), device(NULL) {}
 		OpenCLProgram(OpenCLDevice *device,
 		              const string& program_name,
 		              const string& kernel_name,
@@ -284,24 +280,12 @@ public:
 		~OpenCLProgram();
 
 		void add_kernel(ustring name);
-
-		/* Try to load the program from device cache or disk */
-		bool load();
-		/* Compile the kernel (first separate, failback to local) */
-		void compile();
-		/* Create the OpenCL kernels after loading or compiling */
-		void create_kernels();
+		void load();
 
 		bool is_loaded() const { return loaded; }
 		const string& get_log() const { return log; }
 		void report_error();
 
-		/* Wait until this kernel is available to be used
-		 * It will return true when the kernel is available.
-		 * It will return false when the kernel is not available
-		 * or could not be loaded. */
-		bool wait_for_availability();
-
 		cl_kernel operator()();
 		cl_kernel operator()(ustring name);
 
@@ -325,8 +309,6 @@ public:
 		void add_error(const string& msg);
 
 		bool loaded;
-		bool needs_compiling;
-
 		cl_program program;
 		OpenCLDevice *device;
 
@@ -342,42 +324,26 @@ public:
 		map<ustring, cl_kernel> kernels;
 	};
 
-	/* Container for all types of split programs. */
-	class OpenCLSplitPrograms {
-		public:
-			OpenCLDevice *device;
-			OpenCLProgram program_split;
-			OpenCLProgram program_lamp_emission;
-			OpenCLProgram program_do_volume;
-			OpenCLProgram program_indirect_background;
-			OpenCLProgram program_shader_eval;
-			OpenCLProgram program_holdout_emission_blurring_pathtermination_ao;
-			OpenCLProgram program_subsurface_scatter;
-			OpenCLProgram program_direct_lighting;
-			OpenCLProgram program_shadow_blocked_ao;
-			OpenCLProgram program_shadow_blocked_dl;
-
-			OpenCLSplitPrograms(OpenCLDevice *device);
-			~OpenCLSplitPrograms();
-
-			/* Load the kernels and put the created kernels in the given `programs`
-			 * paramter. */
-			void load_kernels(vector<OpenCLProgram*> &programs,
-			                  const DeviceRequestedFeatures& requested_features,
-			                  bool is_preview=false);
-	};
-
 	DeviceSplitKernel *split_kernel;
 
+	OpenCLProgram program_split;
+
+	OpenCLProgram program_lamp_emission;
+	OpenCLProgram program_do_volume;
+	OpenCLProgram program_indirect_background;
+	OpenCLProgram program_shader_eval;
+	OpenCLProgram program_holdout_emission_blurring_pathtermination_ao;
+	OpenCLProgram program_subsurface_scatter;
+	OpenCLProgram program_direct_lighting;
+	OpenCLProgram program_shadow_blocked_ao;
+	OpenCLProgram program_shadow_blocked_dl;
+
 	OpenCLProgram base_program;
 	OpenCLProgram bake_program;
 	OpenCLProgram displace_program;
 	OpenCLProgram background_program;
 	OpenCLProgram denoising_program;
 
-	OpenCLSplitPrograms kernel_programs;
-	OpenCLSplitPrograms preview_programs;
-
 	typedef map<string, device_vector<uchar>*> ConstMemMap;
 	typedef map<string, device_ptr> MemMap;
 
@@ -393,32 +359,22 @@ public:
 	void opencl_error(const string& message);
 	void opencl_assert_err(cl_int err, const char* where);
 
-	OpenCLDevice(DeviceInfo& info, Stats &stats, Profiler &profiler, bool background);
+	OpenCLDevice(DeviceInfo& info, Stats &stats, Profiler &profiler, bool background_);
 	~OpenCLDevice();
 
 	static void CL_CALLBACK context_notify_callback(const char *err_info,
 		const void * /*private_info*/, size_t /*cb*/, void *user_data);
 
 	bool opencl_version_check();
-	OpenCLSplitPrograms* get_split_programs();
 
 	string device_md5_hash(string kernel_custom_build_options = "");
 	bool load_kernels(const DeviceRequestedFeatures& requested_features);
-	void load_required_kernels(const DeviceRequestedFeatures& requested_features);
-	void load_preview_kernels();
-
-	bool wait_for_availability(const DeviceRequestedFeatures& requested_features);
-	DeviceKernelStatus get_active_kernel_switch_state();
 
 	/* Get the name of the opencl program for the given kernel */
-	const string get_opencl_program_name(const string& kernel_name);
+	const string get_opencl_program_name(bool single_program, const string& kernel_name);
 	/* Get the program file name to compile (*.cl) for the given kernel */
-	const string get_opencl_program_filename(const string& kernel_name);
-	string get_build_options(const DeviceRequestedFeatures& requested_features,
-	                         const string& opencl_program_name,
-	                         bool preview_kernel=false);
-	/* Enable the default features to reduce recompilation events */
-	void enable_default_features(DeviceRequestedFeatures& features);
+	const string get_opencl_program_filename(bool single_program, const string& kernel_name);
+	string get_build_options(const DeviceRequestedFeatures& requested_features, const string& opencl_program_name);
 
 	void mem_alloc(device_memory& mem);
 	void mem_copy_to(device_memory& mem);

intern/cycles/device/opencl/opencl_split.cpp

@@ -40,13 +40,7 @@ struct texture_slot_t {
 	int slot;
 };
 
-static const string NON_SPLIT_KERNELS =
-	"denoising "
-	"base "
-	"background "
-	"displace ";
-
-static const string SPLIT_BUNDLE_KERNELS =
+static const string fast_compiled_kernels =
 	"data_init "
 	"path_init "
 	"state_buffer_size "
@@ -59,52 +53,37 @@ static const string SPLIT_BUNDLE_KERNELS =
 	"indirect_subsurface "
 	"buffer_update";
 
-const string OpenCLDevice::get_opencl_program_name(const string& kernel_name)
+const string OpenCLDevice::get_opencl_program_name(bool single_program, const string& kernel_name)
 {
-	if (NON_SPLIT_KERNELS.find(kernel_name) != std::string::npos) {
-		return kernel_name;
-	}
-	else if (SPLIT_BUNDLE_KERNELS.find(kernel_name) != std::string::npos) {
-		return "split_bundle";
+	if (single_program) {
+		return "split";
 	}
 	else {
-		return "split_" + kernel_name;
+		if (fast_compiled_kernels.find(kernel_name) != std::string::npos) {
+			return "split_bundle";
+		}
+		else {
+			return "split_" + kernel_name;
+		}
 	}
 }
 
-const string OpenCLDevice::get_opencl_program_filename(const string& kernel_name)
+const string OpenCLDevice::get_opencl_program_filename(bool single_program, const string& kernel_name)
 {
-	if (kernel_name == "denoising") {
-		return "filter.cl";
-	}
-	else if (SPLIT_BUNDLE_KERNELS.find(kernel_name) != std::string::npos) {
-		return "kernel_split_bundle.cl";
+	if (single_program) {
+		return "kernel_split.cl";
 	}
 	else {
-		return "kernel_" + kernel_name + ".cl";
+		if (fast_compiled_kernels.find(kernel_name) != std::string::npos) {
+			return "kernel_split_bundle.cl";
+		}
+		else {
+			return "kernel_" + kernel_name + ".cl";
+		}
 	}
 }
 
-/* Enable features that we always want to compile to reduce recompilation events */
-void OpenCLDevice::enable_default_features(DeviceRequestedFeatures& features)
-{
-	features.use_transparent = true;
-	features.use_shadow_tricks = true;
-	features.use_principled = true;
-	features.use_denoising = true;
-
-	if (!background)
-	{
-		features.max_nodes_group = NODE_GROUP_LEVEL_MAX;
-		features.nodes_features = NODE_FEATURE_ALL;
-		features.use_hair = true;
-		features.use_subsurface = true;
-		features.use_camera_motion = false;
-		features.use_object_motion = false;
-	}
-}
-
-string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_features, const string& opencl_program_name, bool preview_kernel)
+string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_features, const string& opencl_program_name)
 {
 	/* first check for non-split kernel programs */
 	if (opencl_program_name == "base" || opencl_program_name == "denoising") {
@@ -112,25 +91,18 @@ string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_
 	}
 	else if (opencl_program_name == "bake") {
 		/* Note: get_build_options for bake is only requested when baking is enabled.
-		 * displace and background are always requested.
-		 * `__SPLIT_KERNEL__` must not be present in the compile directives for bake */
+		   displace and background are always requested.
+		   `__SPLIT_KERNEL__` must not be present in the compile directives for bake */
 		DeviceRequestedFeatures features(requested_features);
-		enable_default_features(features);
 		features.use_denoising = false;
 		features.use_object_motion = false;
 		features.use_camera_motion = false;
-		features.use_hair = true;
-		features.use_subsurface = true;
-		features.max_nodes_group = NODE_GROUP_LEVEL_MAX;
-		features.nodes_features = NODE_FEATURE_ALL;
-		features.use_integrator_branched = false;
 		return features.get_build_options();
 	}
 	else if (opencl_program_name == "displace") {
 		/* As displacement does not use any nodes from the Shading group (eg BSDF).
-		 * We disable all features that are related to shading. */
+		   We disable all features that are related to shading. */
 		DeviceRequestedFeatures features(requested_features);
-		enable_default_features(features);
 		features.use_denoising = false;
 		features.use_object_motion = false;
 		features.use_camera_motion = false;
@@ -142,33 +114,27 @@ string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_
 		features.nodes_features &= ~NODE_FEATURE_VOLUME;
 		features.use_denoising = false;
 		features.use_principled = false;
-		features.use_integrator_branched = false;
 		return features.get_build_options();
 	}
 	else if (opencl_program_name == "background") {
 		/* Background uses Background shading
-		 * It is save to disable shadow features, subsurface and volumetric. */
+		   It is save to disable shadow features, subsurface and volumetric. */
 		DeviceRequestedFeatures features(requested_features);
-		enable_default_features(features);
 		features.use_baking = false;
-		features.use_object_motion = false;
-		features.use_camera_motion = false;
 		features.use_transparent = false;
 		features.use_shadow_tricks = false;
 		features.use_denoising = false;
 		/* NOTE: currently possible to use surface nodes like `Hair Info`, `Bump` node.
-		 * Perhaps we should remove them in UI as it does not make any sense when
-		 * rendering background. */
+		   Perhaps we should remove them in UI as it does not make any sense when
+		   rendering background. */
 		features.nodes_features &= ~NODE_FEATURE_VOLUME;
 		features.use_subsurface = false;
 		features.use_volume = false;
-		features.use_shader_raytrace = false;
-		features.use_patch_evaluation = false;
-		features.use_integrator_branched = false;
 		return features.get_build_options();
 	}
 
 	string build_options = "-D__SPLIT_KERNEL__ ";
+	DeviceRequestedFeatures nofeatures;
 	/* Set compute device build option. */
 	cl_device_type device_type;
 	OpenCLInfo::get_device_type(this->cdDevice, &device_type, &this->ciErr);
@@ -177,34 +143,28 @@ string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_
 		build_options += "-D__COMPUTE_DEVICE_GPU__ ";
 	}
 
-	DeviceRequestedFeatures nofeatures;
-	enable_default_features(nofeatures);
-
 	/* Add program specific optimized compile directives */
-	if (preview_kernel) {
-		DeviceRequestedFeatures preview_features;
-		preview_features.use_hair = true;
-		build_options += "-D__KERNEL_AO_PREVIEW__ ";
-		build_options += preview_features.get_build_options();
+	if (opencl_program_name == "split_do_volume" && !requested_features.use_volume) {
+		build_options += nofeatures.get_build_options();
 	}
-	else if (opencl_program_name == "split_do_volume" && !requested_features.use_volume) {
+	else if (opencl_program_name == "split_subsurface_scatter" && !requested_features.use_subsurface) {
+		/* When subsurface is off, the kernel updates indexes and does not need any
+		   Compile directives */
 		build_options += nofeatures.get_build_options();
 	}
 	else {
 		DeviceRequestedFeatures features(requested_features);
-		enable_default_features(features);
 
 		/* Always turn off baking at this point. Baking is only usefull when building the bake kernel.
-		 * this also makes sure that the kernels that are build during baking can be reused
-		 * when not doing any baking. */
+		   this also makes sure that the kernels that are build during baking can be reused
+		   when not doing any baking. */
 		features.use_baking = false;
 
 		/* Do not vary on shaders when program doesn't do any shading.
-		 * We have bundled them in a single program. */
+		   We have bundled them in a single program. */
 		if (opencl_program_name == "split_bundle") {
 			features.max_nodes_group = 0;
 			features.nodes_features = 0;
-			features.use_shader_raytrace = false;
 		}
 
 		/* No specific settings, just add the regular ones */
@@ -214,77 +174,6 @@ string OpenCLDevice::get_build_options(const DeviceRequestedFeatures& requested_
 	return build_options;
 }
 
-OpenCLDevice::OpenCLSplitPrograms::OpenCLSplitPrograms(OpenCLDevice *device_)
-{
-	device = device_;
-}
-
-OpenCLDevice::OpenCLSplitPrograms::~OpenCLSplitPrograms()
-{
-	program_split.release();
-	program_lamp_emission.release();
-	program_do_volume.release();
-	program_indirect_background.release();
-	program_shader_eval.release();
-	program_holdout_emission_blurring_pathtermination_ao.release();
-	program_subsurface_scatter.release();
-	program_direct_lighting.release();
-	program_shadow_blocked_ao.release();
-	program_shadow_blocked_dl.release();
-}
-
-void OpenCLDevice::OpenCLSplitPrograms::load_kernels(vector<OpenCLProgram*> &programs, const DeviceRequestedFeatures& requested_features, bool is_preview)
-{
-	if (!requested_features.use_baking) {
-#define ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(kernel_name) program_split.add_kernel(ustring("path_trace_"#kernel_name));
-#define ADD_SPLIT_KERNEL_PROGRAM(kernel_name) \
-		const string program_name_##kernel_name = "split_"#kernel_name; \
-		program_##kernel_name = \
-			OpenCLDevice::OpenCLProgram(device, \
-			                            program_name_##kernel_name, \
-			                            "kernel_"#kernel_name".cl", \
-			                            device->get_build_options(requested_features, program_name_##kernel_name, is_preview)); \
-		program_##kernel_name.add_kernel(ustring("path_trace_"#kernel_name)); \
-		programs.push_back(&program_##kernel_name);
-
-		/* Ordered with most complex kernels first, to reduce overall compile time. */
-		ADD_SPLIT_KERNEL_PROGRAM(subsurface_scatter);
-		if (requested_features.use_volume || is_preview) {
-			ADD_SPLIT_KERNEL_PROGRAM(do_volume);
-		}
-		ADD_SPLIT_KERNEL_PROGRAM(shadow_blocked_dl);
-		ADD_SPLIT_KERNEL_PROGRAM(shadow_blocked_ao);
-		ADD_SPLIT_KERNEL_PROGRAM(holdout_emission_blurring_pathtermination_ao);
-		ADD_SPLIT_KERNEL_PROGRAM(lamp_emission);
-		ADD_SPLIT_KERNEL_PROGRAM(direct_lighting);
-		ADD_SPLIT_KERNEL_PROGRAM(indirect_background);
-		ADD_SPLIT_KERNEL_PROGRAM(shader_eval);
-
-		/* Quick kernels bundled in a single program to reduce overhead of starting
-			* Blender processes. */
-		program_split = OpenCLDevice::OpenCLProgram(device,
-		                                            "split_bundle" ,
-		                                            "kernel_split_bundle.cl",
-		                                            device->get_build_options(requested_features, "split_bundle", is_preview));
-
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(data_init);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(state_buffer_size);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(path_init);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(scene_intersect);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(queue_enqueue);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(shader_setup);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(shader_sort);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(enqueue_inactive);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(next_iteration_setup);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(indirect_subsurface);
-		ADD_SPLIT_KERNEL_BUNDLE_PROGRAM(buffer_update);
-		programs.push_back(&program_split);
-
-#undef ADD_SPLIT_KERNEL_PROGRAM
-#undef ADD_SPLIT_KERNEL_BUNDLE_PROGRAM
-	}
-}
-
 namespace {
 
 /* Copy dummy KernelGlobals related to OpenCL from kernel_globals.h to
@@ -391,14 +280,13 @@ public:
 	{
 		OpenCLSplitKernelFunction* kernel = new OpenCLSplitKernelFunction(device, cached_memory);
 
-		const string program_name = device->get_opencl_program_name(kernel_name);
+		bool single_program = OpenCLInfo::use_single_program();
+		const string program_name = device->get_opencl_program_name(single_program, kernel_name);
 		kernel->program =
 			OpenCLDevice::OpenCLProgram(device,
 			                            program_name,
-			                            device->get_opencl_program_filename(kernel_name),
-			                            device->get_build_options(requested_features,
-			                                                      program_name,
-			                                                      device->use_preview_kernels));
+			                            device->get_opencl_program_filename(single_program, kernel_name),
+			                            device->get_build_options(requested_features, program_name));
 
 		kernel->program.add_kernel(ustring("path_trace_" + kernel_name));
 		kernel->program.load();
@@ -418,8 +306,7 @@ public:
 		size_buffer.zero_to_device();
 
 		uint threads = num_threads;
-		OpenCLDevice::OpenCLSplitPrograms *programs = device->get_split_programs();
-		cl_kernel kernel_state_buffer_size = programs->program_split(ustring("path_trace_state_buffer_size"));
+		cl_kernel kernel_state_buffer_size = device->program_split(ustring("path_trace_state_buffer_size"));
 		device->kernel_set_args(kernel_state_buffer_size, 0, kg, data, threads, size_buffer);
 
 		size_t global_size = 64;
@@ -469,8 +356,7 @@ public:
 		cl_int start_sample = rtile.start_sample;
 		cl_int end_sample = rtile.start_sample + rtile.num_samples;
 
-		OpenCLDevice::OpenCLSplitPrograms *programs = device->get_split_programs();
-		cl_kernel kernel_data_init = programs->program_split(ustring("path_trace_data_init"));
+		cl_kernel kernel_data_init = device->program_split(ustring("path_trace_data_init"));
 
 		cl_uint start_arg_index =
 			device->kernel_set_args(kernel_data_init,
@@ -603,8 +489,6 @@ void OpenCLDevice::opencl_assert_err(cl_int err, const char* where)
 
 OpenCLDevice::OpenCLDevice(DeviceInfo& info, Stats &stats, Profiler &profiler, bool background)
 : Device(info, stats, profiler, background),
-  kernel_programs(this),
-  preview_programs(this),
   memory_manager(this),
   texture_info(this, "__texture_info", MEM_TEXTURE)
 {
@@ -615,7 +499,6 @@ OpenCLDevice::OpenCLDevice(DeviceInfo& info, Stats &stats, Profiler &profiler, b
 	null_mem = 0;
 	device_initialized = false;
 	textures_need_update = true;
-	use_preview_kernels = !background;
 
 	vector<OpenCLPlatformDevice> usable_devices;
 	OpenCLInfo::get_usable_devices(&usable_devices);
@@ -679,16 +562,11 @@ OpenCLDevice::OpenCLDevice(DeviceInfo& info, Stats &stats, Profiler &profiler, b
 	device_initialized = true;
 
 	split_kernel = new OpenCLSplitKernel(this);
-	if (!background) {
-		load_preview_kernels();
-	}
 }
 
 OpenCLDevice::~OpenCLDevice()
 {
 	task_pool.stop();
-	load_required_kernel_task_pool.stop();
-	load_kernel_task_pool.stop();
 
 	memory_manager.free();
 
@@ -704,7 +582,7 @@ OpenCLDevice::~OpenCLDevice()
 	bake_program.release();
 	displace_program.release();
 	background_program.release();
-	denoising_program.release();
+	program_split.release();
 
 	if(cqCommandQueue)
 		clReleaseCommandQueue(cqCommandQueue);
@@ -771,50 +649,7 @@ bool OpenCLDevice::load_kernels(const DeviceRequestedFeatures& requested_feature
 	if(!opencl_version_check())
 		return false;
 
-	load_required_kernels(requested_features);
-
 	vector<OpenCLProgram*> programs;
-	kernel_programs.load_kernels(programs, requested_features, false);
-
-	if (!requested_features.use_baking && requested_features.use_denoising) {
-		denoising_program = OpenCLProgram(this, "denoising", "filter.cl", get_build_options(requested_features, "denoising"));
-		denoising_program.add_kernel(ustring("filter_divide_shadow"));
-		denoising_program.add_kernel(ustring("filter_get_feature"));
-		denoising_program.add_kernel(ustring("filter_write_feature"));
-		denoising_program.add_kernel(ustring("filter_detect_outliers"));
-		denoising_program.add_kernel(ustring("filter_combine_halves"));
-		denoising_program.add_kernel(ustring("filter_construct_transform"));
-		denoising_program.add_kernel(ustring("filter_nlm_calc_difference"));
-		denoising_program.add_kernel(ustring("filter_nlm_blur"));
-		denoising_program.add_kernel(ustring("filter_nlm_calc_weight"));
-		denoising_program.add_kernel(ustring("filter_nlm_update_output"));
-		denoising_program.add_kernel(ustring("filter_nlm_normalize"));
-		denoising_program.add_kernel(ustring("filter_nlm_construct_gramian"));
-		denoising_program.add_kernel(ustring("filter_finalize"));
-		programs.push_back(&denoising_program);
-	}
-
-	load_required_kernel_task_pool.wait_work();
-
-	/* Parallel compilation of Cycles kernels, this launches multiple
-	 * processes to workaround OpenCL frameworks serializing the calls
-	 * internally within a single process. */
-	foreach(OpenCLProgram *program, programs) {
-		if (!program->load()) {
-			load_kernel_task_pool.push(function_bind(&OpenCLProgram::compile, program));
-		}
-	}
-	return true;
-}
-
-void OpenCLDevice::load_required_kernels(const DeviceRequestedFeatures& requested_features)
-{
-	vector<OpenCLProgram*> programs;
-	base_program = OpenCLProgram(this, "base", "kernel_base.cl", get_build_options(requested_features, "base"));
-	base_program.add_kernel(ustring("convert_to_byte"));
-	base_program.add_kernel(ustring("convert_to_half_float"));
-	base_program.add_kernel(ustring("zero_buffer"));
-	programs.push_back(&base_program);
 
 	if (requested_features.use_true_displacement) {
 		displace_program = OpenCLProgram(this, "displace", "kernel_displace.cl", get_build_options(requested_features, "displace"));
@@ -828,89 +663,133 @@ void OpenCLDevice::load_required_kernels(const DeviceRequestedFeatures& requeste
 		programs.push_back(&background_program);
 	}
 
+	bool single_program = OpenCLInfo::use_single_program();
+
+#define ADD_SPLIT_KERNEL_SINGLE_PROGRAM(kernel_name) program_split.add_kernel(ustring("path_trace_"#kernel_name));
+#define ADD_SPLIT_KERNEL_SPLIT_PROGRAM(kernel_name) \
+		const string program_name_##kernel_name = "split_"#kernel_name; \
+		program_##kernel_name = \
+			OpenCLDevice::OpenCLProgram(this, \
+			                            program_name_##kernel_name, \
+			                            "kernel_"#kernel_name".cl", \
+			                            get_build_options(requested_features, program_name_##kernel_name)); \
+		program_##kernel_name.add_kernel(ustring("path_trace_"#kernel_name)); \
+		programs.push_back(&program_##kernel_name);
+
+	if (single_program) {
+		program_split = OpenCLDevice::OpenCLProgram(this,
+		                                            "split" ,
+		                                            "kernel_split.cl",
+		                                            get_build_options(requested_features, "split"));
+
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(state_buffer_size);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(data_init);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(path_init);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(scene_intersect);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(lamp_emission);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(do_volume);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(queue_enqueue);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(indirect_background);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shader_setup);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shader_sort);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shader_eval);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(holdout_emission_blurring_pathtermination_ao);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(subsurface_scatter);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(direct_lighting);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shadow_blocked_ao);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shadow_blocked_dl);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(enqueue_inactive);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(next_iteration_setup);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(indirect_subsurface);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(buffer_update);
+
+		programs.push_back(&program_split);
+	}
+	else {
+		/* Ordered with most complex kernels first, to reduce overall compile time. */
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(subsurface_scatter);
+		if (requested_features.use_volume) {
+			ADD_SPLIT_KERNEL_SPLIT_PROGRAM(do_volume);
+		}
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(shadow_blocked_dl);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(shadow_blocked_ao);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(holdout_emission_blurring_pathtermination_ao);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(lamp_emission);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(direct_lighting);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(indirect_background);
+		ADD_SPLIT_KERNEL_SPLIT_PROGRAM(shader_eval);
+
+		/* Quick kernels bundled in a single program to reduce overhead of starting
+			* Blender processes. */
+		program_split = OpenCLDevice::OpenCLProgram(this,
+		                                            "split_bundle" ,
+		                                            "kernel_split_bundle.cl",
+		                                            get_build_options(requested_features, "split_bundle"));
+
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(data_init);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(state_buffer_size);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(path_init);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(scene_intersect);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(queue_enqueue);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shader_setup);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(shader_sort);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(enqueue_inactive);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(next_iteration_setup);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(indirect_subsurface);
+		ADD_SPLIT_KERNEL_SINGLE_PROGRAM(buffer_update);
+		programs.push_back(&program_split);
+	}
+#undef ADD_SPLIT_KERNEL_SPLIT_PROGRAM
+#undef ADD_SPLIT_KERNEL_SINGLE_PROGRAM
+
+	base_program = OpenCLProgram(this, "base", "kernel_base.cl", get_build_options(requested_features, "base"));
+	base_program.add_kernel(ustring("convert_to_byte"));
+	base_program.add_kernel(ustring("convert_to_half_float"));
+	base_program.add_kernel(ustring("zero_buffer"));
+	programs.push_back(&base_program);
+
 	if (requested_features.use_baking) {
 		bake_program = OpenCLProgram(this, "bake", "kernel_bake.cl", get_build_options(requested_features, "bake"));
 		bake_program.add_kernel(ustring("bake"));
 		programs.push_back(&bake_program);
 	}
 
-	foreach(OpenCLProgram *program, programs) {
-		if (!program->load()) {
-			load_required_kernel_task_pool.push(function_bind(&OpenCLProgram::compile, program));
-		}
-	}
-}
+	denoising_program = OpenCLProgram(this, "denoising", "filter.cl", get_build_options(requested_features, "denoising"));
+	denoising_program.add_kernel(ustring("filter_divide_shadow"));
+	denoising_program.add_kernel(ustring("filter_get_feature"));
+	denoising_program.add_kernel(ustring("filter_write_feature"));
+	denoising_program.add_kernel(ustring("filter_detect_outliers"));
+	denoising_program.add_kernel(ustring("filter_combine_halves"));
+	denoising_program.add_kernel(ustring("filter_construct_transform"));
+	denoising_program.add_kernel(ustring("filter_nlm_calc_difference"));
+	denoising_program.add_kernel(ustring("filter_nlm_blur"));
+	denoising_program.add_kernel(ustring("filter_nlm_calc_weight"));
+	denoising_program.add_kernel(ustring("filter_nlm_update_output"));
+	denoising_program.add_kernel(ustring("filter_nlm_normalize"));
+	denoising_program.add_kernel(ustring("filter_nlm_construct_gramian"));
+	denoising_program.add_kernel(ustring("filter_finalize"));
+	programs.push_back(&denoising_program);
 
-void OpenCLDevice::load_preview_kernels()
-{
-	DeviceRequestedFeatures no_features;
-	vector<OpenCLProgram*> programs;
-	preview_programs.load_kernels(programs, no_features, true);
+	/* Parallel compilation of Cycles kernels, this launches multiple
+	 * processes to workaround OpenCL frameworks serializing the calls
+	 * internally within a single process. */
+	TaskPool task_pool;
+	foreach(OpenCLProgram *program, programs) {
+		task_pool.push(function_bind(&OpenCLProgram::load, program));
+	}
+	task_pool.wait_work();
 
 	foreach(OpenCLProgram *program, programs) {
-		if (!program->load()) {
-			load_required_kernel_task_pool.push(function_bind(&OpenCLProgram::compile, program));
+		VLOG(2) << program->get_log();
+		if(!program->is_loaded()) {
+			program->report_error();
+			return false;
 		}
 	}
-}
 
-bool OpenCLDevice::wait_for_availability(const DeviceRequestedFeatures& requested_features)
-{
-	if (background) {
-		load_kernel_task_pool.wait_work();
-		use_preview_kernels = false;
-	}
-	else {
-		/* We use a device setting to determine to load preview kernels or not
-		 * Better to check on device level than per kernel as mixing preview and
-		 * non-preview kernels does not work due to different data types */
-		if (use_preview_kernels) {
-			use_preview_kernels = !load_kernel_task_pool.finished();
-		}
-	}
 	return split_kernel->load_kernels(requested_features);
 }
 
-OpenCLDevice::OpenCLSplitPrograms* OpenCLDevice::get_split_programs()
-{
-	return use_preview_kernels?&preview_programs:&kernel_programs;
-}
-
-DeviceKernelStatus OpenCLDevice::get_active_kernel_switch_state()
-{
-	/* Do not switch kernels for background renderings
-	 * We do foreground rendering but use the preview kernels
-	 * Check for the optimized kernels
-	 *
-	 * This works also the other way around, where we are using
-	 * optimized kernels but new ones are being compiled due
-	 * to other features that are needed */
-	if (background) {
-		/* The if-statements below would find the same result,
-		 * But as the `finished` method uses a mutex we added
-		 * this as an early exit */
-		return DEVICE_KERNEL_USING_FEATURE_KERNEL;
-	}
-
-	bool other_kernels_finished = load_kernel_task_pool.finished();
-	if (use_preview_kernels) {
-		if (other_kernels_finished) {
-			return DEVICE_KERNEL_FEATURE_KERNEL_AVAILABLE;
-		}
-		else {
-			return DEVICE_KERNEL_WAITING_FOR_FEATURE_KERNEL;
-		}
-	}
-	else {
-		if (other_kernels_finished) {
-			return DEVICE_KERNEL_USING_FEATURE_KERNEL;
-		}
-		else {
-			return DEVICE_KERNEL_FEATURE_KERNEL_INVALID;
-		}
-	}
-}
-
 void OpenCLDevice::mem_alloc(device_memory& mem)
 {
 	if(mem.name) {
@@ -1012,7 +891,6 @@ void OpenCLDevice::mem_copy_from(device_memory& mem, int y, int w, int h, int el
 
 void OpenCLDevice::mem_zero_kernel(device_ptr mem, size_t size)
 {
-	base_program.wait_for_availability();
 	cl_kernel ckZeroBuffer = base_program(ustring("zero_buffer"));
 
 	size_t global_size[] = {1024, 1024};
@@ -1352,13 +1230,13 @@ void OpenCLDevice::thread_run(DeviceTask *task)
 
 				/* Complete kernel execution before release tile. */
 				/* This helps in multi-device render;
-				 * The device that reaches the critical-section function
-				 * release_tile waits (stalling other devices from entering
-				 * release_tile) for all kernels to complete. If device1 (a
-				 * slow-render device) reaches release_tile first then it would
-				 * stall device2 (a fast-render device) from proceeding to render
-				 * next tile.
-				 */
+					* The device that reaches the critical-section function
+					* release_tile waits (stalling other devices from entering
+					* release_tile) for all kernels to complete. If device1 (a
+					* slow-render device) reaches release_tile first then it would
+					* stall device2 (a fast-render device) from proceeding to render
+					* next tile.
+					*/
 				clFinish(cqCommandQueue);
 			}
 			else if(tile.task == RenderTile::DENOISE) {
@@ -1840,15 +1718,17 @@ void OpenCLDevice::shader(DeviceTask& task)
 	cl_int d_shader_w = task.shader_w;
 	cl_int d_offset = task.offset;
 
-	OpenCLDevice::OpenCLProgram *program = &background_program;
+	cl_kernel kernel;
+
 	if(task.shader_eval_type >= SHADER_EVAL_BAKE) {
-		program = &bake_program;
+		kernel = bake_program(ustring("bake"));
 	}
 	else if(task.shader_eval_type == SHADER_EVAL_DISPLACE) {
-		program = &displace_program;
+		kernel = displace_program(ustring("displace"));
+	}
+	else {
+		kernel = background_program(ustring("background"));
 	}
-	program->wait_for_availability();
-	cl_kernel kernel = (*program)();
 
 	cl_uint start_arg_index =
 		kernel_set_args(kernel,

intern/cycles/device/opencl/opencl_util.cpp

@@ -243,18 +243,6 @@ string OpenCLCache::get_kernel_md5()
 	return self.kernel_md5;
 }
 
-static string get_program_source(const string& kernel_file)
-{
-	string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\"\n";
-	/* We compile kernels consisting of many files. unfortunately OpenCL
-	 * kernel caches do not seem to recognize changes in included files.
-	 * so we force recompile on changes by adding the md5 hash of all files.
-	 */
-	source = path_source_replace_includes(source, path_get("source"));
-	source += "\n// " + util_md5_string(source) + "\n";
-	return source;
-}
-
 OpenCLDevice::OpenCLProgram::OpenCLProgram(OpenCLDevice *device,
                                                const string& program_name,
                                                const string& kernel_file,
@@ -267,7 +255,6 @@ OpenCLDevice::OpenCLProgram::OpenCLProgram(OpenCLDevice *device,
    use_stdout(use_stdout)
 {
 	loaded = false;
-	needs_compiling = true;
 	program = NULL;
 }
 
@@ -356,7 +343,13 @@ bool OpenCLDevice::OpenCLProgram::build_kernel(const string *debug_src)
 
 bool OpenCLDevice::OpenCLProgram::compile_kernel(const string *debug_src)
 {
-	string source = get_program_source(kernel_file);
+	string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\"\n";
+	/* We compile kernels consisting of many files. unfortunately OpenCL
+	 * kernel caches do not seem to recognize changes in included files.
+	 * so we force recompile on changes by adding the md5 hash of all files.
+	 */
+	source = path_source_replace_includes(source, path_get("source"));
+	source += "\n// " + util_md5_string(source) + "\n";
 
 	if(debug_src) {
 		path_write_text(*debug_src, source);
@@ -480,7 +473,8 @@ bool device_opencl_compile_kernel(const vector<string>& parameters)
 		return false;
 	}
 
-	string source = get_program_source(kernel_file);
+	string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\" // " + path_files_md5_hash(path_get("kernel")) + "\n";
+	source = path_source_replace_includes(source, path_get("source"));
 	size_t source_len = source.size();
 	const char *source_str = source.c_str();
 	cl_program program = clCreateProgramWithSource(context, 1, &source_str, &source_len, &err);
@@ -554,55 +548,12 @@ bool OpenCLDevice::OpenCLProgram::save_binary(const string& clbin)
 	return path_write_binary(clbin, binary);
 }
 
-bool OpenCLDevice::OpenCLProgram::load()
-{
-	loaded = false;
-	string device_md5 = device->device_md5_hash(kernel_build_options);
-
-	/* Try to use cached kernel. */
-	thread_scoped_lock cache_locker;
-	ustring cache_key(program_name + device_md5);
-	program = device->load_cached_kernel(cache_key,
-	                                     cache_locker);
-	if (!program) {
-		add_log(string("OpenCL program ") + program_name + " not found in cache.", true);
-
-		/* need to create source to get md5 */
-		string source = get_program_source(kernel_file);
-
-		string basename = "cycles_kernel_" + program_name + "_" + device_md5 + "_" + util_md5_string(source);
-		basename = path_cache_get(path_join("kernels", basename));
-		string clbin = basename + ".clbin";
-
-		/* If binary kernel exists already, try use it. */
-		if(path_exists(clbin) && load_binary(clbin)) {
-			/* Kernel loaded from binary, nothing to do. */
-			add_log(string("Loaded program from ") + clbin + ".", true);
-
-			/* Cache the program. */
-			device->store_cached_kernel(program,
-			                            cache_key,
-			                            cache_locker);
-		}
-		else {
-			add_log(string("OpenCL program ") + program_name + " not found on disk.", true);
-			cache_locker.unlock();
-		}
-	}
-
-	if (program) {
-		create_kernels();
-		loaded = true;
-		needs_compiling = false;
-	}
-
-	return loaded;
-}
-
-void OpenCLDevice::OpenCLProgram::compile()
+void OpenCLDevice::OpenCLProgram::load()
 {
 	assert(device);
 
+	loaded = false;
+
 	string device_md5 = device->device_md5_hash(kernel_build_options);
 
 	/* Try to use cached kernel. */
@@ -611,13 +562,12 @@ void OpenCLDevice::OpenCLProgram::compile()
 	program = device->load_cached_kernel(cache_key,
 	                                     cache_locker);
 
-	if (!program)
-	{
-
+	if(!program) {
 		add_log(string("OpenCL program ") + program_name + " not found in cache.", true);
 
 		/* need to create source to get md5 */
-		string source = get_program_source(kernel_file);
+		string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\"\n";
+		source = path_source_replace_includes(source, path_get("source"));
 
 		string basename = "cycles_kernel_" + program_name + "_" + device_md5 + "_" + util_md5_string(source);
 		basename = path_cache_get(path_join("kernels", basename));
@@ -632,38 +582,49 @@ void OpenCLDevice::OpenCLProgram::compile()
 		}
 
 		/* If binary kernel exists already, try use it. */
-		if(compile_separate(clbin)) {
-			add_log(string("Built and loaded program from ") + clbin + ".", true);
-			loaded = true;
+		if(path_exists(clbin) && load_binary(clbin)) {
+			/* Kernel loaded from binary, nothing to do. */
+			add_log(string("Loaded program from ") + clbin + ".", true);
 		}
 		else {
-			add_log(string("Separate-process building of ") + clbin + " failed, will fall back to regular building.", true);
+			add_log(string("Kernel file ") + clbin + " either doesn't exist or failed to be loaded by driver.", true);
+			if(!path_exists(clbin)) {
+				if(compile_separate(clbin)) {
+					add_log(string("Built and loaded program from ") + clbin + ".", true);
+					loaded = true;
+				}
+				else {
+					add_log(string("Separate-process building of ") + clbin + " failed, will fall back to regular building.", true);
 
-			/* If does not exist or loading binary failed, compile kernel. */
-			if(!compile_kernel(debug_src)) {
-				needs_compiling = false;
-				return;
+					/* If does not exist or loading binary failed, compile kernel. */
+					if(!compile_kernel(debug_src)) {
+						return;
+					}
+
+					/* Save binary for reuse. */
+					if(!save_binary(clbin)) {
+						add_log(string("Saving compiled OpenCL kernel to ") + clbin + " failed!", true);
+					}
+				}
 			}
-
-			/* Save binary for reuse. */
-			if(!save_binary(clbin)) {
-				add_log(string("Saving compiled OpenCL kernel to ") + clbin + " failed!", true);
+			else {
+				add_log(string("Kernel file ") + clbin + "exists, but failed to be loaded by driver.", true);
+				/* Fall back to compiling. */
+				if(!compile_kernel(debug_src)) {
+					return;
+				}
 			}
 		}
 
 		/* Cache the program. */
 		device->store_cached_kernel(program,
-									cache_key,
-									cache_locker);
+		                            cache_key,
+		                            cache_locker);
+	}
+	else {
+		add_log(string("Found cached OpenCL program ") + program_name + ".", true);
 	}
 
-	create_kernels();
-	needs_compiling = false;
-	loaded = true;
-}
-
-void OpenCLDevice::OpenCLProgram::create_kernels()
-{
 	for(map<ustring, cl_kernel>::iterator kernel = kernels.begin(); kernel != kernels.end(); ++kernel) {
 		assert(kernel->second == NULL);
 		cl_int ciErr;
@@ -674,15 +635,8 @@ void OpenCLDevice::OpenCLProgram::create_kernels()
 			return;
 		}
 	}
-}
 
-bool OpenCLDevice::OpenCLProgram::wait_for_availability()
-{
-	add_log(string("Waiting for availability of ") + program_name + ".", true);
-	while (needs_compiling) {
-		time_sleep(0.1);
-	}
-	return loaded;
+	loaded = true;
 }
 
 void OpenCLDevice::OpenCLProgram::report_error()
@@ -737,6 +691,28 @@ bool OpenCLInfo::use_debug()
 	return DebugFlags().opencl.debug;
 }
 
+bool OpenCLInfo::use_single_program()
+{
+	return DebugFlags().opencl.single_program;
+}
+
+bool OpenCLInfo::kernel_use_advanced_shading(const string& platform)
+{
+	/* keep this in sync with kernel_types.h! */
+	if(platform == "NVIDIA CUDA")
+		return true;
+	else if(platform == "Apple")
+		return true;
+	else if(platform == "AMD Accelerated Parallel Processing")
+		return true;
+	else if(platform == "Intel(R) OpenCL")
+		return true;
+	/* Make sure officially unsupported OpenCL platforms
+	 * does not set up to use advanced shading.
+	 */
+	return false;
+}
+
 bool OpenCLInfo::device_supported(const string& platform_name,
                                   const cl_device_id device_id)
 {

intern/cycles/kernel/CMakeLists.txt

@@ -41,6 +41,7 @@ set(SRC_OPENCL_KERNELS
 	kernels/opencl/kernel_displace.cl
 	kernels/opencl/kernel_background.cl
 	kernels/opencl/kernel_state_buffer_size.cl
+	kernels/opencl/kernel_split.cl
 	kernels/opencl/kernel_split_bundle.cl
 	kernels/opencl/kernel_data_init.cl
 	kernels/opencl/kernel_path_init.cl
@@ -346,11 +347,11 @@ if(WITH_CYCLES_CUDA_BINARIES)
 	set(CUDA_VERSION "${CUDA_VERSION_MAJOR}${CUDA_VERSION_MINOR}")
 
 	# warn for other versions
-	if(CUDA_VERSION MATCHES "101")
+	if(CUDA_VERSION MATCHES "90" OR CUDA_VERSION MATCHES "91" OR CUDA_VERSION MATCHES "100")
 	else()
 		message(WARNING
 			"CUDA version ${CUDA_VERSION_MAJOR}.${CUDA_VERSION_MINOR} detected, "
-			"build may succeed but only CUDA 10.1 is officially supported")
+			"build may succeed but only CUDA 9.0, 9.1 and 10.0 are officially supported")
 	endif()
 
 	# build for each arch
@@ -400,17 +401,29 @@ if(WITH_CYCLES_CUDA_BINARIES)
 			set(cuda_flags ${cuda_flags} -D __KERNEL_DEBUG__)
 		endif()
 
-		if(WITH_CYCLES_CUBIN_COMPILER)
+		# Workaround to build only sm_7x kernels with CUDA 10, until
+		# older kernels work well with this version.
+		if(DEFINED CUDA10_NVCC_EXECUTABLE  AND (${arch} MATCHES "sm_7."))
+			set(with_cubin_compiler OFF)
+			set(cuda_nvcc_executable "${CUDA10_NVCC_EXECUTABLE}")
+			set(cuda_toolkit_root_dir "${CUDA10_TOOLKIT_ROOT_DIR}")
+		else()
+			set(with_cubin_compiler ${WITH_CYCLES_CUBIN_COMPILER})
+			set(cuda_nvcc_executable "${CUDA_NVCC_EXECUTABLE}")
+			set(cuda_toolkit_root_dir "${CUDA_TOOLKIT_ROOT_DIR}")
+		endif()
+
+		if(with_cubin_compiler)
 			string(SUBSTRING ${arch} 3 -1 CUDA_ARCH)
 
 			# Needed to find libnvrtc-builtins.so. Can't do it from inside
 			# cycles_cubin_cc since the env variable is read before main()
 			if(APPLE)
 				set(CUBIN_CC_ENV ${CMAKE_COMMAND}
-					-E env DYLD_LIBRARY_PATH="${CUDA_TOOLKIT_ROOT_DIR}/lib")
+					-E env DYLD_LIBRARY_PATH="${cuda_toolkit_root_dir}/lib")
 			elseif(UNIX)
 				set(CUBIN_CC_ENV ${CMAKE_COMMAND}
-					-E env LD_LIBRARY_PATH="${CUDA_TOOLKIT_ROOT_DIR}/lib64")
+					-E env LD_LIBRARY_PATH="${cuda_toolkit_root_dir}/lib64")
 			endif()
 
 			add_custom_command(
@@ -421,12 +434,12 @@ if(WITH_CYCLES_CUDA_BINARIES)
 						-i ${CMAKE_CURRENT_SOURCE_DIR}${cuda_kernel_src}
 						${cuda_flags}
 						-v
-						-cuda-toolkit-dir "${CUDA_TOOLKIT_ROOT_DIR}"
+						-cuda-toolkit-dir "${cuda_toolkit_root_dir}"
 				DEPENDS ${kernel_sources} cycles_cubin_cc)
 		else()
 			add_custom_command(
 				OUTPUT ${cuda_cubin}
-				COMMAND ${CUDA_NVCC_EXECUTABLE}
+				COMMAND ${cuda_nvcc_executable}
 						-arch=${arch}
 						${CUDA_NVCC_FLAGS}
 						--cubin
@@ -445,7 +458,7 @@ if(WITH_CYCLES_CUDA_BINARIES)
 	foreach(arch ${CYCLES_CUDA_BINARIES_ARCH})
 		if(${arch} MATCHES "sm_2.")
 			message(STATUS "CUDA binaries for ${arch} are no longer supported, skipped.")
-		elseif(${arch} MATCHES "sm_7." AND ${CUDA_VERSION} LESS 100)
+		elseif(${arch} MATCHES "sm_7." AND (${CUDA_VERSION} LESS 100) AND (NOT DEFINED CUDA10_NVCC_EXECUTABLE))
 			message(STATUS "CUDA binaries for ${arch} require CUDA 10.0+, skipped.")
 		else()
 			# Compile regular kernel

intern/cycles/kernel/bvh/bvh.h

@@ -253,7 +253,6 @@ ccl_device_intersect bool scene_intersect_local(KernelGlobals *kg,
 	PROFILING_INIT(kg, PROFILING_INTERSECT_LOCAL);
 
 	if(!scene_intersect_valid(&ray)) {
-		local_isect->num_hits = 0;
 		return false;
 	}
 #ifdef __EMBREE__

intern/cycles/kernel/filter/filter_reconstruction.h

@@ -95,16 +95,14 @@ ccl_device_inline void kernel_filter_finalize(int x, int y,
 	}
 
 	/* The weighted average of pixel colors (essentially, the NLM-filtered image).
-	 * In case the solution of the linear model fails due to numerical issues or
-	 * returns non-sensical negative values, fall back to this value. */
+	 * In case the solution of the linear model fails due to numerical issues,
+	 * fall back to this value. */
 	float3 mean_color = XtWY[0]/XtWX[0];
 
 	math_trimatrix_vec3_solve(XtWX, XtWY, (*rank)+1, stride);
 
 	float3 final_color = XtWY[0];
-	if(!isfinite3_safe(final_color) ||
-	   (final_color.x < -0.01f || final_color.y < -0.01f || final_color.z < -0.01f))
-	{
+	if(!isfinite3_safe(final_color)) {
 		final_color = mean_color;
 	}
 

intern/cycles/kernel/geom/geom_curve.h

@@ -87,45 +87,6 @@ ccl_device float curve_attribute_float(KernelGlobals *kg, const ShaderData *sd,
 	}
 }
 
-ccl_device float2 curve_attribute_float2(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float2 *dx, float2 *dy)
-{
-	if(desc.element == ATTR_ELEMENT_CURVE) {
-		/* idea: we can't derive any useful differentials here, but for tiled
-		 * mipmap image caching it would be useful to avoid reading the highest
-		 * detail level always. maybe a derivative based on the hair density
-		 * could be computed somehow? */
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-#endif
-
-		return kernel_tex_fetch(__attributes_float2, desc.offset + sd->prim);
-	}
-	else if(desc.element == ATTR_ELEMENT_CURVE_KEY || desc.element == ATTR_ELEMENT_CURVE_KEY_MOTION) {
-		float4 curvedata = kernel_tex_fetch(__curves, sd->prim);
-		int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type);
-		int k1 = k0 + 1;
-
-		float2 f0 = kernel_tex_fetch(__attributes_float2, desc.offset + k0);
-		float2 f1 = kernel_tex_fetch(__attributes_float2, desc.offset + k1);
-
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx) *dx = sd->du.dx*(f1 - f0);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-#endif
-
-		return (1.0f - sd->u)*f0 + sd->u*f1;
-	}
-	else {
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-#endif
-
-		return make_float2(0.0f, 0.0f);
-	}
-}
-
 ccl_device float3 curve_attribute_float3(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float3 *dx, float3 *dy)
 {
 	if(desc.element == ATTR_ELEMENT_CURVE) {

intern/cycles/kernel/geom/geom_patch.h

@@ -284,33 +284,6 @@ ccl_device float patch_eval_float(KernelGlobals *kg, const ShaderData *sd, int o
 	return val;
 }
 
-ccl_device float2 patch_eval_float2(KernelGlobals *kg, const ShaderData *sd, int offset,
-                                    int patch, float u, float v, int channel,
-                                    float2 *du, float2 *dv)
-{
-	int indices[PATCH_MAX_CONTROL_VERTS];
-	float weights[PATCH_MAX_CONTROL_VERTS];
-	float weights_du[PATCH_MAX_CONTROL_VERTS];
-	float weights_dv[PATCH_MAX_CONTROL_VERTS];
-
-	int num_control = patch_eval_control_verts(kg, sd->object, patch, u, v, channel,
-	                                           indices, weights, weights_du, weights_dv);
-
-	float2 val = make_float2(0.0f, 0.0f);
-	if(du) *du = make_float2(0.0f, 0.0f);
-	if(dv) *dv = make_float2(0.0f, 0.0f);
-
-	for(int i = 0; i < num_control; i++) {
-		float2 v = kernel_tex_fetch(__attributes_float2, offset + indices[i]);
-
-		val += v * weights[i];
-		if(du) *du += v * weights_du[i];
-		if(dv) *dv += v * weights_dv[i];
-	}
-
-	return val;
-}
-
 ccl_device float3 patch_eval_float3(KernelGlobals *kg, const ShaderData *sd, int offset,
                                     int patch, float u, float v, int channel,
                                     float3 *du, float3 *dv)

intern/cycles/kernel/geom/geom_primitive.h

@@ -89,37 +89,6 @@ ccl_device_inline float primitive_volume_attribute_float(KernelGlobals *kg,
 }
 #endif
 
-ccl_device_inline float2 primitive_attribute_float2(KernelGlobals *kg,
-                                                    const ShaderData *sd,
-                                                    const AttributeDescriptor desc,
-                                                    float2 *dx, float2 *dy)
-{
-	if(sd->type & PRIMITIVE_ALL_TRIANGLE) {
-		if(subd_triangle_patch(kg, sd) == ~0)
-			return triangle_attribute_float2(kg, sd, desc, dx, dy);
-		else
-			return subd_triangle_attribute_float2(kg, sd, desc, dx, dy);
-	}
-#ifdef __HAIR__
-	else if(sd->type & PRIMITIVE_ALL_CURVE) {
-		return curve_attribute_float2(kg, sd, desc, dx, dy);
-	}
-#endif
-#ifdef __VOLUME__
-	else if(sd->object != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL) {
-		kernel_assert(0);
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-		return make_float2(0.0f, 0.0f);
-	}
-#endif
-	else {
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-		return make_float2(0.0f, 0.0f);
-	}
-}
-
 ccl_device_inline float3 primitive_attribute_float3(KernelGlobals *kg,
                                                     const ShaderData *sd,
                                                     const AttributeDescriptor desc,
@@ -150,29 +119,6 @@ ccl_device_inline float3 primitive_attribute_float3(KernelGlobals *kg,
 	}
 }
 
-ccl_device_inline float2 primitive_surface_attribute_float2(KernelGlobals *kg,
-                                                            const ShaderData *sd,
-                                                            const AttributeDescriptor desc,
-                                                            float2 *dx, float2 *dy)
-{
-	if(sd->type & PRIMITIVE_ALL_TRIANGLE) {
-		if(subd_triangle_patch(kg, sd) == ~0)
-			return triangle_attribute_float2(kg, sd, desc, dx, dy);
-		else
-			return subd_triangle_attribute_float2(kg, sd, desc, dx, dy);
-	}
-#ifdef __HAIR__
-	else if(sd->type & PRIMITIVE_ALL_CURVE) {
-		return curve_attribute_float2(kg, sd, desc, dx, dy);
-	}
-#endif
-	else {
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-		return make_float2(0.0f, 0.0f);
-	}
-}
-
 ccl_device_inline float3 primitive_surface_attribute_float3(KernelGlobals *kg,
                                                             const ShaderData *sd,
                                                             const AttributeDescriptor desc,
@@ -219,8 +165,9 @@ ccl_device_inline float3 primitive_uv(KernelGlobals *kg, ShaderData *sd)
 	if(desc.offset == ATTR_STD_NOT_FOUND)
 		return make_float3(0.0f, 0.0f, 0.0f);
 
-	float2 uv = primitive_surface_attribute_float2(kg, sd, desc, NULL, NULL);
-	return make_float3(uv.x, uv.y, 1.0f);
+	float3 uv = primitive_surface_attribute_float3(kg, sd, desc, NULL, NULL);
+	uv.z = 1.0f;
+	return uv;
 }
 
 /* Ptex coordinates */

intern/cycles/kernel/geom/geom_subd_triangle.h

@@ -216,128 +216,6 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const
 	}
 }
 
-ccl_device_noinline float2 subd_triangle_attribute_float2(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float2 *dx, float2 *dy)
-{
-	int patch = subd_triangle_patch(kg, sd);
-
-#ifdef __PATCH_EVAL__
-	if(desc.flags & ATTR_SUBDIVIDED) {
-		float2 uv[3];
-		subd_triangle_patch_uv(kg, sd, uv);
-
-		float2 dpdu = uv[0] - uv[2];
-		float2 dpdv = uv[1] - uv[2];
-
-		/* p is [s, t] */
-		float2 p = dpdu * sd->u + dpdv * sd->v + uv[2];
-
-		float2 a, dads, dadt;
-
-		a = patch_eval_float2(kg, sd, desc.offset, patch, p.x, p.y, 0, &dads, &dadt);
-
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx || dy) {
-			float dsdu = dpdu.x;
-			float dtdu = dpdu.y;
-			float dsdv = dpdv.x;
-			float dtdv = dpdv.y;
-
-			if(dx) {
-				float dudx = sd->du.dx;
-				float dvdx = sd->dv.dx;
-
-				float dsdx = dsdu*dudx + dsdv*dvdx;
-				float dtdx = dtdu*dudx + dtdv*dvdx;
-
-				*dx = dads*dsdx + dadt*dtdx;
-			}
-			if(dy) {
-				float dudy = sd->du.dy;
-				float dvdy = sd->dv.dy;
-
-				float dsdy = dsdu*dudy + dsdv*dvdy;
-				float dtdy = dtdu*dudy + dtdv*dvdy;
-
-				*dy = dads*dsdy + dadt*dtdy;
-			}
-		}
-#endif
-
-		return a;
-	}
-	else
-#endif  /* __PATCH_EVAL__ */
-		if(desc.element == ATTR_ELEMENT_FACE) {
-			if(dx) *dx = make_float2(0.0f, 0.0f);
-			if(dy) *dy = make_float2(0.0f, 0.0f);
-
-			return kernel_tex_fetch(__attributes_float2, desc.offset + subd_triangle_patch_face(kg, patch));
-		}
-		else if(desc.element == ATTR_ELEMENT_VERTEX || desc.element == ATTR_ELEMENT_VERTEX_MOTION) {
-			float2 uv[3];
-			subd_triangle_patch_uv(kg, sd, uv);
-
-			uint4 v = subd_triangle_patch_indices(kg, patch);
-
-			float2 f0 = kernel_tex_fetch(__attributes_float2, desc.offset + v.x);
-			float2 f1 = kernel_tex_fetch(__attributes_float2, desc.offset + v.y);
-			float2 f2 = kernel_tex_fetch(__attributes_float2, desc.offset + v.z);
-			float2 f3 = kernel_tex_fetch(__attributes_float2, desc.offset + v.w);
-
-			if(subd_triangle_patch_num_corners(kg, patch) != 4) {
-				f1 = (f1+f0)*0.5f;
-				f3 = (f3+f0)*0.5f;
-			}
-
-			float2 a = mix(mix(f0, f1, uv[0].x), mix(f3, f2, uv[0].x), uv[0].y);
-			float2 b = mix(mix(f0, f1, uv[1].x), mix(f3, f2, uv[1].x), uv[1].y);
-			float2 c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y);
-
-#ifdef __RAY_DIFFERENTIALS__
-			if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c;
-			if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c;
-#endif
-
-			return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c;
-		}
-		else if(desc.element == ATTR_ELEMENT_CORNER) {
-			float2 uv[3];
-			subd_triangle_patch_uv(kg, sd, uv);
-
-			int corners[4];
-			subd_triangle_patch_corners(kg, patch, corners);
-
-			float2 f0, f1, f2, f3;
-
-			f0 = kernel_tex_fetch(__attributes_float2, corners[0] + desc.offset);
-			f1 = kernel_tex_fetch(__attributes_float2, corners[1] + desc.offset);
-			f2 = kernel_tex_fetch(__attributes_float2, corners[2] + desc.offset);
-			f3 = kernel_tex_fetch(__attributes_float2, corners[3] + desc.offset);
-
-			if(subd_triangle_patch_num_corners(kg, patch) != 4) {
-				f1 = (f1+f0)*0.5f;
-				f3 = (f3+f0)*0.5f;
-			}
-
-			float2 a = mix(mix(f0, f1, uv[0].x), mix(f3, f2, uv[0].x), uv[0].y);
-			float2 b = mix(mix(f0, f1, uv[1].x), mix(f3, f2, uv[1].x), uv[1].y);
-			float2 c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y);
-
-#ifdef __RAY_DIFFERENTIALS__
-			if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c;
-			if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c;
-#endif
-
-			return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c;
-		}
-		else {
-			if(dx) *dx = make_float2(0.0f, 0.0f);
-			if(dy) *dy = make_float2(0.0f, 0.0f);
-
-			return make_float2(0.0f, 0.0f);
-		}
-}
-
 ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float3 *dx, float3 *dy)
 {
 	int patch = subd_triangle_patch(kg, sd);

intern/cycles/kernel/geom/geom_triangle.h

@@ -149,53 +149,6 @@ ccl_device float triangle_attribute_float(KernelGlobals *kg, const ShaderData *s
 	}
 }
 
-ccl_device float2 triangle_attribute_float2(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float2 *dx, float2 *dy)
-{
-	if(desc.element == ATTR_ELEMENT_FACE) {
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-
-		return kernel_tex_fetch(__attributes_float2, desc.offset + sd->prim);
-	}
-	else if(desc.element == ATTR_ELEMENT_VERTEX || desc.element == ATTR_ELEMENT_VERTEX_MOTION) {
-		uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim);
-
-		float2 f0 = kernel_tex_fetch(__attributes_float2, desc.offset + tri_vindex.x);
-		float2 f1 = kernel_tex_fetch(__attributes_float2, desc.offset + tri_vindex.y);
-		float2 f2 = kernel_tex_fetch(__attributes_float2, desc.offset + tri_vindex.z);
-
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2;
-		if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2;
-#endif
-
-		return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2;
-	}
-	else if(desc.element == ATTR_ELEMENT_CORNER) {
-		int tri = desc.offset + sd->prim*3;
-		float2 f0, f1, f2;
-
-		if(desc.element == ATTR_ELEMENT_CORNER) {
-			f0 = kernel_tex_fetch(__attributes_float2, tri + 0);
-			f1 = kernel_tex_fetch(__attributes_float2, tri + 1);
-			f2 = kernel_tex_fetch(__attributes_float2, tri + 2);
-		}
-
-#ifdef __RAY_DIFFERENTIALS__
-		if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2;
-		if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2;
-#endif
-
-		return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2;
-	}
-	else {
-		if(dx) *dx = make_float2(0.0f, 0.0f);
-		if(dy) *dy = make_float2(0.0f, 0.0f);
-
-		return make_float2(0.0f, 0.0f);
-	}
-}
-
 ccl_device float3 triangle_attribute_float3(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float3 *dx, float3 *dy)
 {
 	if(desc.element == ATTR_ELEMENT_FACE) {

intern/cycles/kernel/kernel_bake.h

@@ -351,7 +351,7 @@ ccl_device void kernel_bake_evaluate(KernelGlobals *kg, ccl_global uint4 *input,
 				out = make_float3(roughness, roughness, roughness);
 			}
 			else {
-				out = shader_emissive_eval(&sd);
+				out = shader_emissive_eval(kg, &sd);
 			}
 			break;
 		}
@@ -475,9 +475,8 @@ ccl_device void kernel_bake_evaluate(KernelGlobals *kg, ccl_global uint4 *input,
 			shader_setup_from_background(kg, &sd, &ray);
 
 			/* evaluate */
-			int path_flag = 0; /* we can't know which type of BSDF this is for */
-			shader_eval_surface(kg, &sd, &state, path_flag | PATH_RAY_EMISSION);
-			out = shader_background_eval(&sd);
+			int flag = 0; /* we can't know which type of BSDF this is for */
+			out = shader_eval_background(kg, &sd, &state, flag);
 			break;
 		}
 		default:
@@ -555,9 +554,8 @@ ccl_device void kernel_background_evaluate(KernelGlobals *kg,
 	shader_setup_from_background(kg, &sd, &ray);
 
 	/* evaluate */
-	int path_flag = 0; /* we can't know which type of BSDF this is for */
-	shader_eval_surface(kg, &sd, &state, path_flag | PATH_RAY_EMISSION);
-	float3 color = shader_background_eval(&sd);
+	int flag = 0; /* we can't know which type of BSDF this is for */
+	float3 color = shader_eval_background(kg, &sd, &state, flag);
 
 	/* write output */
 	output[i] += make_float4(color.x, color.y, color.z, 0.0f);

intern/cycles/kernel/kernel_compat_opencl.h

@@ -125,9 +125,7 @@
 #define fmodf(x, y) fmod((float)(x), (float)(y))
 #define sinhf(x) sinh(((float)(x)))
 
-/* Use native functions with possibly lower precision for performance,
- * no issues found so far. */
-#if 1
+#ifndef __CL_USE_NATIVE__
 #  define sinf(x) native_sin(((float)(x)))
 #  define cosf(x) native_cos(((float)(x)))
 #  define tanf(x) native_tan(((float)(x)))
@@ -142,7 +140,7 @@
 #  define expf(x) exp(((float)(x)))
 #  define sqrtf(x) sqrt(((float)(x)))
 #  define logf(x) log(((float)(x)))
-#  define rcp(x)  recip(x)
+#  define rcp(x)  recip(x))
 #endif
 
 /* data lookup defines */

intern/cycles/kernel/kernel_emission.h

@@ -29,36 +29,43 @@ ccl_device_noinline float3 direct_emissive_eval(KernelGlobals *kg,
 	/* setup shading at emitter */
 	float3 eval;
 
-	if(shader_constant_emission_eval(kg, ls->shader, &eval)) {
+	int shader_flag = kernel_tex_fetch(__shaders, (ls->shader & SHADER_MASK)).flags;
+
+#ifdef __BACKGROUND_MIS__
+	if(ls->type == LIGHT_BACKGROUND) {
+		Ray ray;
+		ray.D = ls->D;
+		ray.P = ls->P;
+		ray.t = 1.0f;
+		ray.time = time;
+		ray.dP = differential3_zero();
+		ray.dD = dI;
+
+		shader_setup_from_background(kg, emission_sd, &ray);
+
+		path_state_modify_bounce(state, true);
+		eval = shader_eval_background(kg, emission_sd, state, 0);
+		path_state_modify_bounce(state, false);
+	}
+	else
+#endif
+	if(shader_flag & SD_HAS_CONSTANT_EMISSION)
+	{
+		eval.x = kernel_tex_fetch(__shaders, (ls->shader & SHADER_MASK)).constant_emission[0];
+		eval.y = kernel_tex_fetch(__shaders, (ls->shader & SHADER_MASK)).constant_emission[1];
+		eval.z = kernel_tex_fetch(__shaders, (ls->shader & SHADER_MASK)).constant_emission[2];
 		if((ls->prim != PRIM_NONE) && dot(ls->Ng, I) < 0.0f) {
 			ls->Ng = -ls->Ng;
 		}
 	}
-	else {
-		/* Setup shader data and call shader_eval_surface once, better
-		 * for GPU coherence and compile times. */
-#ifdef __BACKGROUND_MIS__
-		if(ls->type == LIGHT_BACKGROUND) {
-			Ray ray;
-			ray.D = ls->D;
-			ray.P = ls->P;
-			ray.t = 1.0f;
-			ray.time = time;
-			ray.dP = differential3_zero();
-			ray.dD = dI;
+	else
+	{
+		shader_setup_from_sample(kg, emission_sd,
+		                         ls->P, ls->Ng, I,
+		                         ls->shader, ls->object, ls->prim,
+		                         ls->u, ls->v, t, time, false, ls->lamp);
 
-			shader_setup_from_background(kg, emission_sd, &ray);
-		}
-		else
-#endif
-		{
-			shader_setup_from_sample(kg, emission_sd,
-			                         ls->P, ls->Ng, I,
-			                         ls->shader, ls->object, ls->prim,
-			                         ls->u, ls->v, t, time, false, ls->lamp);
-
-			ls->Ng = emission_sd->Ng;
-		}
+		ls->Ng = emission_sd->Ng;
 
 		/* No proper path flag, we're evaluating this for all closures. that's
 		 * weak but we'd have to do multiple evaluations otherwise. */
@@ -66,16 +73,8 @@ ccl_device_noinline float3 direct_emissive_eval(KernelGlobals *kg,
 		shader_eval_surface(kg, emission_sd, state, PATH_RAY_EMISSION);
 		path_state_modify_bounce(state, false);
 
-		/* Evaluate closures. */
-#ifdef __BACKGROUND_MIS__
-		if (ls->type == LIGHT_BACKGROUND) {
-			eval = shader_background_eval(emission_sd);
-		}
-		else
-#endif
-		{
-			eval = shader_emissive_eval(emission_sd);
-		}
+		/* Evaluate emissive closure. */
+		eval = shader_emissive_eval(kg, emission_sd);
 	}
 
 	eval *= ls->eval_fac;
@@ -202,7 +201,7 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg,
 ccl_device_noinline float3 indirect_primitive_emission(KernelGlobals *kg, ShaderData *sd, float t, int path_flag, float bsdf_pdf)
 {
 	/* evaluate emissive closure */
-	float3 L = shader_emissive_eval(sd);
+	float3 L = shader_emissive_eval(kg, sd);
 
 #ifdef __HAIR__
 	if(!(path_flag & PATH_RAY_MIS_SKIP) && (sd->flag & SD_USE_MIS) && (sd->type & PRIMITIVE_ALL_TRIANGLE))
@@ -295,7 +294,7 @@ ccl_device_noinline float3 indirect_background(KernelGlobals *kg,
 #ifdef __BACKGROUND__
 	int shader = kernel_data.background.surface_shader;
 
-	/* Use visibility flag to skip lights. */
+	/* use visibility flag to skip lights */
 	if(shader & SHADER_EXCLUDE_ANY) {
 		if(((shader & SHADER_EXCLUDE_DIFFUSE) && (state->flag & PATH_RAY_DIFFUSE)) ||
 		   ((shader & SHADER_EXCLUDE_GLOSSY) &&
@@ -306,27 +305,20 @@ ccl_device_noinline float3 indirect_background(KernelGlobals *kg,
 			return make_float3(0.0f, 0.0f, 0.0f);
 	}
 
-
-	/* Evaluate background shader. */
-	float3 L;
-	if(!shader_constant_emission_eval(kg, shader, &L)) {
+	/* evaluate background closure */
 #  ifdef __SPLIT_KERNEL__
-		Ray priv_ray = *ray;
-		shader_setup_from_background(kg, emission_sd, &priv_ray);
+	Ray priv_ray = *ray;
+	shader_setup_from_background(kg, emission_sd, &priv_ray);
 #  else
-		shader_setup_from_background(kg, emission_sd, ray);
+	shader_setup_from_background(kg, emission_sd, ray);
 #  endif
 
-		path_state_modify_bounce(state, true);
-		shader_eval_surface(kg, emission_sd, state, state->flag | PATH_RAY_EMISSION);
-		path_state_modify_bounce(state, false);
+	path_state_modify_bounce(state, true);
+	float3 L = shader_eval_background(kg, emission_sd, state, state->flag);
+	path_state_modify_bounce(state, false);
 
-		L = shader_background_eval(emission_sd);
-	}
-
-	/* Background MIS weights. */
 #ifdef __BACKGROUND_MIS__
-	/* Check if background light exists or if we should skip pdf. */
+	/* check if background light exists or if we should skip pdf */
 	int res_x = kernel_data.integrator.pdf_background_res_x;
 
 	if(!(state->flag & PATH_RAY_MIS_SKIP) && res_x) {

intern/cycles/kernel/kernel_shader.h

@@ -56,7 +56,7 @@ ccl_device_noinline void shader_setup_from_ray(KernelGlobals *kg,
 	PROFILING_INIT(kg, PROFILING_SHADER_SETUP);
 
 #ifdef __INSTANCING__
-	sd->object = (isect->object == OBJECT_NONE)? kernel_tex_fetch(__prim_object, isect->prim): isect->object;
+	sd->object = (isect->object == PRIM_NONE)? kernel_tex_fetch(__prim_object, isect->prim): isect->object;
 #endif
 	sd->lamp = LAMP_NONE;
 
@@ -299,10 +299,6 @@ ccl_device_inline void shader_setup_from_sample(KernelGlobals *kg,
 		sd->ob_tfm  = lamp_fetch_transform(kg, lamp, false);
 		sd->ob_itfm = lamp_fetch_transform(kg, lamp, true);
 		sd->lamp = lamp;
-#else
-	}
-	else if(lamp != LAMP_NONE) {
-		sd->lamp = lamp;
 #endif
 	}
 
@@ -411,7 +407,7 @@ ccl_device_inline void shader_setup_from_background(KernelGlobals *kg, ShaderDat
 	sd->ray_length = 0.0f;
 
 #ifdef __INSTANCING__
-	sd->object = OBJECT_NONE;
+	sd->object = PRIM_NONE;
 #endif
 	sd->lamp = LAMP_NONE;
 	sd->prim = PRIM_NONE;
@@ -434,9 +430,6 @@ ccl_device_inline void shader_setup_from_background(KernelGlobals *kg, ShaderDat
 	sd->dv = differential_zero();
 #endif
 
-	/* for NDC coordinates */
-	sd->ray_P = ray->P;
-
 	PROFILING_SHADER(sd->shader);
 	PROFILING_OBJECT(sd->object);
 }
@@ -460,7 +453,7 @@ ccl_device_inline void shader_setup_from_volume(KernelGlobals *kg, ShaderData *s
 	sd->ray_length = 0.0f; /* todo: can we set this to some useful value? */
 
 #  ifdef __INSTANCING__
-	sd->object = OBJECT_NONE; /* todo: fill this for texture coordinates */
+	sd->object = PRIM_NONE; /* todo: fill this for texture coordinates */
 #  endif
 	sd->lamp = LAMP_NONE;
 	sd->prim = PRIM_NONE;
@@ -987,40 +980,9 @@ ccl_device float3 shader_bssrdf_sum(ShaderData *sd, float3 *N_, float *texture_b
 }
 #endif  /* __SUBSURFACE__ */
 
-/* Constant emission optimization */
-
-ccl_device bool shader_constant_emission_eval(KernelGlobals *kg, int shader, float3 *eval)
-{
-	int shader_index = shader & SHADER_MASK;
-	int shader_flag = kernel_tex_fetch(__shaders, shader_index).flags;
-
-	if (shader_flag & SD_HAS_CONSTANT_EMISSION) {
-		*eval = make_float3(
-			kernel_tex_fetch(__shaders, shader_index).constant_emission[0],
-			kernel_tex_fetch(__shaders, shader_index).constant_emission[1],
-			kernel_tex_fetch(__shaders, shader_index).constant_emission[2]);
-
-		return true;
-	}
-
-	return false;
-}
-
-/* Background */
-
-ccl_device float3 shader_background_eval(ShaderData *sd)
-{
-	if(sd->flag & SD_EMISSION) {
-		return sd->closure_emission_background;
-	}
-	else {
-		return make_float3(0.0f, 0.0f, 0.0f);
-	}
-}
-
 /* Emission */
 
-ccl_device float3 shader_emissive_eval(ShaderData *sd)
+ccl_device float3 shader_emissive_eval(KernelGlobals *kg, ShaderData *sd)
 {
 	if(sd->flag & SD_EMISSION) {
 		return emissive_simple_eval(sd->Ng, sd->I) * sd->closure_emission_background;
@@ -1068,32 +1030,20 @@ ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd,
 	sd->num_closure_left = max_closures;
 
 #ifdef __OSL__
-	if(kg->osl) {
-		if (sd->object == OBJECT_NONE) {
-			OSLShader::eval_background(kg, sd, state, path_flag);
-		}
-		else {
-			OSLShader::eval_surface(kg, sd, state, path_flag);
-		}
-	}
+	if(kg->osl)
+		OSLShader::eval_surface(kg, sd, state, path_flag);
 	else
 #endif
 	{
 #ifdef __SVM__
 		svm_eval_nodes(kg, sd, state, SHADER_TYPE_SURFACE, path_flag);
 #else
-		if(sd->object == OBJECT_NONE) {
-			sd->closure_emission_background = make_float3(0.8f, 0.8f, 0.8f);
-			sd->flag |= SD_EMISSION;
-		}
-		else {
-			DiffuseBsdf *bsdf = (DiffuseBsdf*)bsdf_alloc(sd,
-			                                             sizeof(DiffuseBsdf),
-			                                             make_float3(0.8f, 0.8f, 0.8f));
-			if(bsdf != NULL) {
-				bsdf->N = sd->N;
-				sd->flag |= bsdf_diffuse_setup(bsdf);
-			}
+		DiffuseBsdf *bsdf = (DiffuseBsdf*)bsdf_alloc(sd,
+		                                             sizeof(DiffuseBsdf),
+		                                             make_float3(0.8f, 0.8f, 0.8f));
+		if(bsdf != NULL) {
+			bsdf->N = sd->N;
+			sd->flag |= bsdf_diffuse_setup(bsdf);
 		}
 #endif
 	}
@@ -1103,6 +1053,36 @@ ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd,
 	}
 }
 
+/* Background Evaluation */
+
+ccl_device float3 shader_eval_background(KernelGlobals *kg, ShaderData *sd,
+	ccl_addr_space PathState *state, int path_flag)
+{
+	sd->num_closure = 0;
+	sd->num_closure_left = 0;
+
+#ifdef __SVM__
+#  ifdef __OSL__
+	if(kg->osl) {
+		OSLShader::eval_background(kg, sd, state, path_flag);
+	}
+	else
+#  endif  /* __OSL__ */
+	{
+		svm_eval_nodes(kg, sd, state, SHADER_TYPE_SURFACE, path_flag);
+	}
+
+	if(sd->flag & SD_EMISSION) {
+		return sd->closure_emission_background;
+	}
+	else {
+		return make_float3(0.0f, 0.0f, 0.0f);
+	}
+#else  /* __SVM__ */
+	return make_float3(0.8f, 0.8f, 0.8f);
+#endif  /* __SVM__ */
+}
+
 /* Volume */
 
 #ifdef __VOLUME__

intern/cycles/kernel/kernel_textures.h

@@ -56,7 +56,6 @@ KERNEL_TEX(uint, __patches)
 /* attributes */
 KERNEL_TEX(uint4, __attributes_map)
 KERNEL_TEX(float, __attributes_float)
-KERNEL_TEX(float2, __attributes_float2)
 KERNEL_TEX(float4, __attributes_float3)
 KERNEL_TEX(uchar4, __attributes_uchar4)
 

intern/cycles/kernel/kernel_types.h

@@ -83,6 +83,92 @@ CCL_NAMESPACE_BEGIN
 #  define SHADER_SORT_LOCAL_SIZE 1
 #endif
 
+
+/* Device capabilities */
+#ifdef __KERNEL_CPU__
+#  ifdef __KERNEL_SSE2__
+#    define __QBVH__
+#  endif
+#  define __KERNEL_SHADING__
+#  define __KERNEL_ADV_SHADING__
+#  define __BRANCHED_PATH__
+#  ifdef WITH_OSL
+#    define __OSL__
+#  endif
+#  define __PRINCIPLED__
+#  define __SUBSURFACE__
+#  define __CMJ__
+#  define __VOLUME__
+#  define __VOLUME_SCATTER__
+#  define __SHADOW_RECORD_ALL__
+#  define __VOLUME_DECOUPLED__
+#  define __VOLUME_RECORD_ALL__
+#endif  /* __KERNEL_CPU__ */
+
+#ifdef __KERNEL_CUDA__
+#  define __KERNEL_SHADING__
+#  define __KERNEL_ADV_SHADING__
+#  define __VOLUME__
+#  define __VOLUME_SCATTER__
+#  define __SUBSURFACE__
+#  define __PRINCIPLED__
+#  define __SHADOW_RECORD_ALL__
+#  define __CMJ__
+#  ifndef __SPLIT_KERNEL__
+#    define __BRANCHED_PATH__
+#  endif
+#endif  /* __KERNEL_CUDA__ */
+
+#ifdef __KERNEL_OPENCL__
+
+/* keep __KERNEL_ADV_SHADING__ in sync with opencl_kernel_use_advanced_shading! */
+
+#  ifdef __KERNEL_OPENCL_NVIDIA__
+#    define __KERNEL_SHADING__
+#    define __KERNEL_ADV_SHADING__
+#    define __SUBSURFACE__
+#    define __PRINCIPLED__
+#    define __VOLUME__
+#    define __VOLUME_SCATTER__
+#    define __SHADOW_RECORD_ALL__
+#    define __CMJ__
+#    define __BRANCHED_PATH__
+#  endif  /* __KERNEL_OPENCL_NVIDIA__ */
+
+#  ifdef __KERNEL_OPENCL_APPLE__
+#    define __KERNEL_SHADING__
+#    define __KERNEL_ADV_SHADING__
+#    define __PRINCIPLED__
+#    define __CMJ__
+/* TODO(sergey): Currently experimental section is ignored here,
+ * this is because megakernel in device_opencl does not support
+ * custom cflags depending on the scene features.
+ */
+#  endif  /* __KERNEL_OPENCL_APPLE__ */
+
+#  ifdef __KERNEL_OPENCL_AMD__
+#    define __CL_USE_NATIVE__
+#    define __KERNEL_SHADING__
+#    define __KERNEL_ADV_SHADING__
+#    define __SUBSURFACE__
+#    define __PRINCIPLED__
+#    define __VOLUME__
+#    define __VOLUME_SCATTER__
+#    define __SHADOW_RECORD_ALL__
+#    define __CMJ__
+#    define __BRANCHED_PATH__
+#  endif  /* __KERNEL_OPENCL_AMD__ */
+
+#  ifdef __KERNEL_OPENCL_INTEL_CPU__
+#    define __CL_USE_NATIVE__
+#    define __KERNEL_SHADING__
+#    define __KERNEL_ADV_SHADING__
+#    define __PRINCIPLED__
+#    define __CMJ__
+#  endif  /* __KERNEL_OPENCL_INTEL_CPU__ */
+
+#endif  /* __KERNEL_OPENCL__ */
+
 /* Kernel features */
 #define __SOBOL__
 #define __INSTANCING__
@@ -99,55 +185,28 @@ CCL_NAMESPACE_BEGIN
 #define __SHADOW_TRICKS__
 #define __DENOISING_FEATURES__
 #define __SHADER_RAYTRACE__
-#define __AO__
-#define __PASSES__
-#define __HAIR__
 
-/* Without these we get an AO render, used by OpenCL preview kernel. */
-#ifndef __KERNEL_AO_PREVIEW__
+#ifdef __KERNEL_SHADING__
 #  define __SVM__
 #  define __EMISSION__
 #  define __TEXTURES__
 #  define __EXTRA_NODES__
 #  define __HOLDOUT__
+#endif
+
+#ifdef __KERNEL_ADV_SHADING__
 #  define __MULTI_CLOSURE__
 #  define __TRANSPARENT_SHADOWS__
+#  define __PASSES__
 #  define __BACKGROUND_MIS__
 #  define __LAMP_MIS__
+#  define __AO__
 #  define __CAMERA_MOTION__
 #  define __OBJECT_MOTION__
 #  define __HAIR__
 #  define __BAKING__
-#  define __PRINCIPLED__
-#  define __SUBSURFACE__
-#  define __VOLUME__
-#  define __VOLUME_SCATTER__
-#  define __CMJ__
-#  define __SHADOW_RECORD_ALL__
-#  define __BRANCHED_PATH__
 #endif
 
-/* Device specific features */
-#ifdef __KERNEL_CPU__
-#  ifdef __KERNEL_SSE2__
-#    define __QBVH__
-#  endif
-#  ifdef WITH_OSL
-#    define __OSL__
-#  endif
-#  define __VOLUME_DECOUPLED__
-#  define __VOLUME_RECORD_ALL__
-#endif  /* __KERNEL_CPU__ */
-
-#ifdef __KERNEL_CUDA__
-#  ifdef __SPLIT_KERNEL__
-#    undef __BRANCHED_PATH__
-#  endif
-#endif  /* __KERNEL_CUDA__ */
-
-#ifdef __KERNEL_OPENCL__
-#endif  /* __KERNEL_OPENCL__ */
-
 /* Scene-based selective features compilation. */
 #ifdef __NO_CAMERA_MOTION__
 #  undef __CAMERA_MOTION__

intern/cycles/kernel/kernel_work_stealing.h

@@ -66,15 +66,9 @@ ccl_device_inline void get_work_pixel(ccl_global const WorkTile *tile,
                                       ccl_private uint *y,
                                       ccl_private uint *sample)
 {
-#ifdef __KERNEL_CUDA__
-	/* Keeping threads for the same pixel together improves performance on CUDA. */
-	uint sample_offset = global_work_index % tile->num_samples;
-	uint pixel_offset = global_work_index / tile->num_samples;
-#else /* __KERNEL_CUDA__ */
 	uint tile_pixels = tile->w * tile->h;
 	uint sample_offset = global_work_index / tile_pixels;
 	uint pixel_offset = global_work_index - sample_offset * tile_pixels;
-#endif /* __KERNEL_CUDA__ */
 	uint y_offset = pixel_offset / tile->w;
 	uint x_offset = pixel_offset - y_offset * tile->w;
 

intern/cycles/kernel/kernels/opencl/kernel_split.cl

@@ -0,0 +1,41 @@
+/*
+ * Copyright 2011-2017 Blender Foundation
+ *
+ * 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 "kernel/kernel_compat_opencl.h"  // PRECOMPILED
+#include "kernel/split/kernel_split_common.h"  // PRECOMPILED
+
+#include "kernel/kernels/opencl/kernel_state_buffer_size.cl"
+#include "kernel/kernels/opencl/kernel_data_init.cl"
+#include "kernel/kernels/opencl/kernel_path_init.cl"
+
+#include "kernel/kernels/opencl/kernel_scene_intersect.cl"
+#include "kernel/kernels/opencl/kernel_lamp_emission.cl"
+#include "kernel/kernels/opencl/kernel_do_volume.cl"
+#include "kernel/kernels/opencl/kernel_indirect_background.cl"
+#include "kernel/kernels/opencl/kernel_queue_enqueue.cl"
+#include "kernel/kernels/opencl/kernel_shader_setup.cl"
+#include "kernel/kernels/opencl/kernel_shader_sort.cl"
+#include "kernel/kernels/opencl/kernel_shader_eval.cl"
+#include "kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl"
+#include "kernel/kernels/opencl/kernel_subsurface_scatter.cl"
+#include "kernel/kernels/opencl/kernel_direct_lighting.cl"
+#include "kernel/kernels/opencl/kernel_shadow_blocked_ao.cl"
+#include "kernel/kernels/opencl/kernel_shadow_blocked_dl.cl"
+#include "kernel/kernels/opencl/kernel_enqueue_inactive.cl"
+#include "kernel/kernels/opencl/kernel_next_iteration_setup.cl"
+#include "kernel/kernels/opencl/kernel_indirect_subsurface.cl"
+#include "kernel/kernels/opencl/kernel_buffer_update.cl"
+

intern/cycles/kernel/osl/osl_services.cpp

@@ -392,44 +392,6 @@ bool OSLRenderServices::get_array_attribute(OSL::ShaderGlobals *sg, bool derivat
 	return false;
 }
 
-static bool set_attribute_float2(float2 f[3], TypeDesc type, bool derivatives, void *val)
-{
-	if(type == TypeDesc::TypePoint || type == TypeDesc::TypeVector ||
-	   type == TypeDesc::TypeNormal || type == TypeDesc::TypeColor)
-	{
-		float *fval = (float *)val;
-
-		fval[0] = f[0].x;
-		fval[1] = f[0].y;
-		fval[2] = 0.0f;
-
-		if(derivatives) {
-			fval[3] = f[1].x;
-			fval[4] = f[1].y;
-			fval[5] = 0.0f;
-
-			fval[6] = f[2].x;
-			fval[7] = f[2].y;
-			fval[8] = 0.0f;
-		}
-
-		return true;
-	}
-	else if(type == TypeDesc::TypeFloat) {
-		float *fval = (float *)val;
-		fval[0] = average(f[0]);
-
-		if(derivatives) {
-			fval[1] = average(f[1]);
-			fval[2] = average(f[2]);
-		}
-
-		return true;
-	}
-
-	return false;
-}
-
 static bool set_attribute_float3(float3 f[3], TypeDesc type, bool derivatives, void *val)
 {
 	if(type == TypeDesc::TypePoint || type == TypeDesc::TypeVector ||
@@ -610,12 +572,6 @@ static bool get_primitive_attribute(KernelGlobals *kg, const ShaderData *sd, con
 		                                     (derivatives) ? &fval[1] : NULL, (derivatives) ? &fval[2] : NULL);
 		return set_attribute_float3(fval, type, derivatives, val);
 	}
-	else if(attr.type == TypeFloat2) {
-		float2 fval[2];
-		fval[0] = primitive_attribute_float2(kg, sd, attr.desc,
-		                                      (derivatives) ? &fval[1] : NULL, (derivatives) ? &fval[2] : NULL);
-		return set_attribute_float2(fval, type, derivatives, val);
-	}
 	else if(attr.type == TypeDesc::TypeFloat) {
 		float fval[3];
 		fval[0] = primitive_attribute_float(kg, sd, attr.desc,

intern/cycles/kernel/svm/svm_attribute.h

@@ -52,27 +52,18 @@ ccl_device void svm_node_attr(KernelGlobals *kg, ShaderData *sd, float *stack, u
 	AttributeDescriptor desc = svm_node_attr_init(kg, sd, node, &type, &out_offset);
 
 	/* fetch and store attribute */
-	if(desc.type == NODE_ATTR_FLOAT) {
+	if (desc.type == NODE_ATTR_FLOAT) {
 		float f = primitive_attribute_float(kg, sd, desc, NULL, NULL);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, f);
 		}
 		else {
 			stack_store_float3(stack, out_offset, make_float3(f, f, f));
 		}
 	}
-	else if(desc.type == NODE_ATTR_FLOAT2) {
-		float2 f = primitive_attribute_float2(kg, sd, desc, NULL, NULL);
-		if(type == NODE_ATTR_FLOAT) {
-			stack_store_float(stack, out_offset, f.x);
-		}
-		else {
-			stack_store_float3(stack, out_offset, make_float3(f.x, f.y, 0.0f));
-		}
-	}
 	else {
 		float3 f = primitive_attribute_float3(kg, sd, desc, NULL, NULL);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, average(f));
 		}
 		else {
@@ -93,30 +84,20 @@ void svm_node_attr_bump_dx(KernelGlobals *kg, ShaderData *sd, float *stack, uint
 	AttributeDescriptor desc = svm_node_attr_init(kg, sd, node, &type, &out_offset);
 
 	/* fetch and store attribute */
-	if(desc.type == NODE_ATTR_FLOAT) {
+	if (desc.type == NODE_ATTR_FLOAT) {
 		float dx;
 		float f = primitive_surface_attribute_float(kg, sd, desc, &dx, NULL);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, f+dx);
 		}
 		else {
 			stack_store_float3(stack, out_offset, make_float3(f+dx, f+dx, f+dx));
 		}
 	}
-	else if(desc.type == NODE_ATTR_FLOAT2) {
-		float2 dx;
-		float2 f = primitive_attribute_float2(kg, sd, desc, &dx, NULL);
-		if (type == NODE_ATTR_FLOAT) {
-			stack_store_float(stack, out_offset, f.x + dx.x);
-		}
-		else {
-			stack_store_float3(stack, out_offset, make_float3(f.x+dx.x, f.y+dx.y, 0.0f));
-		}
-	}
 	else {
 		float3 dx;
 		float3 f = primitive_surface_attribute_float3(kg, sd, desc, &dx, NULL);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, average(f+dx));
 		}
 		else {
@@ -140,30 +121,20 @@ void svm_node_attr_bump_dy(KernelGlobals *kg,
 	AttributeDescriptor desc = svm_node_attr_init(kg, sd, node, &type, &out_offset);
 
 	/* fetch and store attribute */
-	if(desc.type == NODE_ATTR_FLOAT) {
+		if (desc.type == NODE_ATTR_FLOAT) {
 		float dy;
 		float f = primitive_surface_attribute_float(kg, sd, desc, NULL, &dy);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, f+dy);
 		}
 		else {
 			stack_store_float3(stack, out_offset, make_float3(f+dy, f+dy, f+dy));
 		}
 	}
-	else if(desc.type == NODE_ATTR_FLOAT2) {
-		float2 dy;
-		float2 f = primitive_attribute_float2(kg, sd, desc, NULL, &dy);
-		if(type == NODE_ATTR_FLOAT) {
-			stack_store_float(stack, out_offset, f.x + dy.x);
-		}
-		else {
-			stack_store_float3(stack, out_offset, make_float3(f.x+dy.x, f.y+dy.y, 0.0f));
-		}
-	}
 	else {
 		float3 dy;
 		float3 f = primitive_surface_attribute_float3(kg, sd, desc, NULL, &dy);
-		if(type == NODE_ATTR_FLOAT) {
+		if (type == NODE_ATTR_FLOAT) {
 			stack_store_float(stack, out_offset, average(f+dy));
 		}
 		else {

intern/cycles/kernel/svm/svm_bevel.h

@@ -146,13 +146,7 @@ ccl_device_noinline float3 svm_bevel(
 			}
 #endif  /* __OBJECT_MOTION__ */
 
-			/* Get geometric normal. */
 			float3 hit_Ng = isect.Ng[hit];
-			int object = (isect.hits[hit].object == OBJECT_NONE)? kernel_tex_fetch(__prim_object, isect.hits[hit].prim): isect.hits[hit].object;
-			int object_flag = kernel_tex_fetch(__object_flag, object);
-			if(object_flag & SD_OBJECT_NEGATIVE_SCALE_APPLIED) {
-				hit_Ng = -hit_Ng;
-			}
 
 			/* Compute smooth normal. */
 			float3 N = hit_Ng;
@@ -174,7 +168,7 @@ ccl_device_noinline float3 svm_bevel(
 			}
 
 			/* Transform normals to world space. */
-			if(!(object_flag & SD_OBJECT_TRANSFORM_APPLIED)) {
+			if(isect.hits[hit].object != OBJECT_NONE) {
 				object_normal_transform(kg, sd, &N);
 				object_normal_transform(kg, sd, &hit_Ng);
 			}

intern/cycles/kernel/svm/svm_tex_coord.h

@@ -363,15 +363,7 @@ ccl_device void svm_node_tangent(KernelGlobals *kg, ShaderData *sd, float *stack
 	float3 attribute_value;
 	const AttributeDescriptor desc = find_attribute(kg, sd, node.z);
 	if (desc.offset != ATTR_STD_NOT_FOUND) {
-		if(desc.type == NODE_ATTR_FLOAT2) {
-			float2 value = primitive_surface_attribute_float2(kg, sd, desc, NULL, NULL);
-			attribute_value.x = value.x;
-			attribute_value.y = value.y;
-			attribute_value.z = 0.0f;
-		}
-		else {
-			attribute_value = primitive_surface_attribute_float3(kg, sd, desc, NULL, NULL);
-		}
+		attribute_value = primitive_surface_attribute_float3(kg, sd, desc, NULL, NULL);
 	}
 
 

intern/cycles/kernel/svm/svm_types.h

@@ -141,7 +141,6 @@ typedef enum ShaderNodeType {
 
 typedef enum NodeAttributeType {
 	NODE_ATTR_FLOAT = 0,
-	NODE_ATTR_FLOAT2,
 	NODE_ATTR_FLOAT3,
 	NODE_ATTR_MATRIX
 } NodeAttributeType;

intern/cycles/render/CMakeLists.txt

@@ -22,7 +22,6 @@ set(SRC
 	image.cpp
 	integrator.cpp
 	light.cpp
-	merge.cpp
 	mesh.cpp
 	mesh_displace.cpp
 	mesh_subdivision.cpp
@@ -56,7 +55,6 @@ set(SRC_HEADERS
 	image.h
 	integrator.h
 	light.h
-	merge.h
 	mesh.h
 	nodes.h
 	object.h

intern/cycles/render/attribute.cpp

@@ -48,8 +48,7 @@ void Attribute::set(ustring name_, TypeDesc type_, AttributeElement element_)
 	/* string and matrix not supported! */
 	assert(type == TypeDesc::TypeFloat || type == TypeDesc::TypeColor ||
 		type == TypeDesc::TypePoint || type == TypeDesc::TypeVector ||
-		type == TypeDesc::TypeNormal || type == TypeDesc::TypeMatrix ||
-		type == TypeFloat2);
+		type == TypeDesc::TypeNormal || type == TypeDesc::TypeMatrix);
 }
 
 void Attribute::resize(Mesh *mesh, AttributePrimitive prim, bool reserve_only)
@@ -69,8 +68,6 @@ void Attribute::resize(size_t num_elements)
 
 void Attribute::add(const float& f)
 {
-	assert(data_sizeof() == sizeof(float));
-
 	char *data = (char*)&f;
 	size_t size = sizeof(f);
 
@@ -80,19 +77,6 @@ void Attribute::add(const float& f)
 
 void Attribute::add(const uchar4& f)
 {
-	assert(data_sizeof() == sizeof(uchar4));
-
-	char *data = (char*)&f;
-	size_t size = sizeof(f);
-
-	for(size_t i = 0; i < size; i++)
-		buffer.push_back(data[i]);
-}
-
-void Attribute::add(const float2& f)
-{
-	assert(data_sizeof() == sizeof(float2));
-
 	char *data = (char*)&f;
 	size_t size = sizeof(f);
 
@@ -102,8 +86,6 @@ void Attribute::add(const float2& f)
 
 void Attribute::add(const float3& f)
 {
-	assert(data_sizeof() == sizeof(float3));
-
 	char *data = (char*)&f;
 	size_t size = sizeof(f);
 
@@ -113,8 +95,6 @@ void Attribute::add(const float3& f)
 
 void Attribute::add(const Transform& f)
 {
-	assert(data_sizeof() == sizeof(Transform));
-
 	char *data = (char*)&f;
 	size_t size = sizeof(f);
 
@@ -124,8 +104,6 @@ void Attribute::add(const Transform& f)
 
 void Attribute::add(const VoxelAttribute& f)
 {
-	assert(data_sizeof() == sizeof(VoxelAttribute));
-
 	char *data = (char*)&f;
 	size_t size = sizeof(f);
 
@@ -149,8 +127,6 @@ size_t Attribute::data_sizeof() const
 		return sizeof(uchar4);
 	else if(type == TypeDesc::TypeFloat)
 		return sizeof(float);
-	else if(type == TypeFloat2)
-		return sizeof(float2);
 	else if(type == TypeDesc::TypeMatrix)
 		return sizeof(Transform);
 	else
@@ -424,7 +400,7 @@ Attribute *AttributeSet::add(AttributeStandard std, ustring name)
 				attr = add(name, TypeDesc::TypeNormal, ATTR_ELEMENT_FACE);
 				break;
 			case ATTR_STD_UV:
-				attr = add(name, TypeFloat2, ATTR_ELEMENT_CORNER);
+				attr = add(name, TypeDesc::TypePoint, ATTR_ELEMENT_CORNER);
 				break;
 			case ATTR_STD_UV_TANGENT:
 				attr = add(name, TypeDesc::TypeVector, ATTR_ELEMENT_CORNER);
@@ -475,8 +451,6 @@ Attribute *AttributeSet::add(AttributeStandard std, ustring name)
 	else if(curve_mesh) {
 		switch(std) {
 			case ATTR_STD_UV:
-				attr = add(name, TypeFloat2, ATTR_ELEMENT_CURVE);
-				break;
 			case ATTR_STD_GENERATED:
 				attr = add(name, TypeDesc::TypePoint, ATTR_ELEMENT_CURVE);
 				break;

intern/cycles/render/attribute.h

@@ -66,86 +66,25 @@ public:
 	size_t element_size(Mesh *mesh, AttributePrimitive prim) const;
 	size_t buffer_size(Mesh *mesh, AttributePrimitive prim) const;
 
-	char *data()
-	{
-		return (buffer.size())? &buffer[0]: NULL;
-	}
-	float2 *data_float2()
-	{
-		assert(data_sizeof() == sizeof(float2));
-		return (float2*)data();
-	}
-	float3 *data_float3()
-	{
-		assert(data_sizeof() == sizeof(float3));
-		return (float3*)data();
-	}
-	float4 *data_float4()
-	{
-		assert(data_sizeof() == sizeof(float4));
-		return (float4*)data();
-	}
-	float *data_float()
-	{
-		assert(data_sizeof() == sizeof(float));
-		return (float*)data();
-	}
-	uchar4 *data_uchar4()
-	{
-		assert(data_sizeof() == sizeof(uchar4));
-		return (uchar4*)data();
-	}
-	Transform *data_transform()
-	{
-		assert(data_sizeof() == sizeof(Transform));
-		return (Transform*)data();
-	}
-	VoxelAttribute *data_voxel()
-	{
-		assert(data_sizeof() == sizeof(VoxelAttribute));
-		return ( VoxelAttribute*)data();
-	}
+	char *data() { return (buffer.size())? &buffer[0]: NULL; };
+	float3 *data_float3() { return (float3*)data(); }
+	float4 *data_float4() { return (float4*)data(); }
+	float *data_float() { return (float*)data(); }
+	uchar4 *data_uchar4() { return (uchar4*)data(); }
+	Transform *data_transform() { return (Transform*)data(); }
+	VoxelAttribute *data_voxel()  { return ( VoxelAttribute*)data(); }
 
-	const char *data() const
-	{
-		return (buffer.size())? &buffer[0]: NULL;
-	}
-	const float2 *data_float2() const
-	{
-		assert(data_sizeof() == sizeof(float2));
-		return (const float2*)data();
-	}
-	const float3 *data_float3() const
-	{
-		assert(data_sizeof() == sizeof(float3));
-		return (const float3*)data();
-	}
-	const float4 *data_float4() const
-	{
-		assert(data_sizeof() == sizeof(float4));
-		return (const float4*)data();
-	}
-	const float *data_float() const
-	{
-		assert(data_sizeof() == sizeof(float));
-		return (const float*)data();
-	}
-	const Transform *data_transform() const
-	{
-		assert(data_sizeof() == sizeof(Transform));
-		return (const Transform*)data();
-	}
-	const VoxelAttribute *data_voxel() const
-	{
-		assert(data_sizeof() == sizeof(VoxelAttribute));
-		return (const VoxelAttribute*)data();
-	}
+	const char *data() const { return (buffer.size())? &buffer[0]: NULL; }
+	const float3 *data_float3() const { return (const float3*)data(); }
+	const float4 *data_float4() const { return (const float4*)data(); }
+	const float *data_float() const { return (const float*)data(); }
+	const Transform *data_transform() const { return (const Transform*)data(); }
+	const VoxelAttribute *data_voxel() const { return (const VoxelAttribute*)data(); }
 
 	void zero_data(void* dst);
 	void add_with_weight(void* dst, void* src, float weight);
 
 	void add(const float& f);
-	void add(const float2& f);
 	void add(const float3& f);
 	void add(const uchar4& f);
 	void add(const Transform& f);

intern/cycles/render/buffers.cpp

@@ -184,7 +184,6 @@ bool RenderBuffers::get_denoising_pass_rect(int type, float exposure, int sample
 	if(type == DENOISING_PASS_CLEAN) {
 		/* The clean pass isn't changed by prefiltering, so we use the original one there. */
 		offset = type + params.get_denoising_offset();
-		scale /= sample;
 	}
 	else if (type == DENOISING_PASS_PREFILTERED_COLOR && !params.denoising_prefiltered_pass) {
 		/* If we're not saving the prefiltering result, return the original noisy pass. */

intern/cycles/render/camera.cpp

@@ -169,8 +169,6 @@ Camera::Camera()
 	cameratoworld = transform_identity();
 	worldtoraster = projection_identity();
 
-	full_rastertocamera = projection_identity();
-
 	dx = make_float3(0.0f, 0.0f, 0.0f);
 	dy = make_float3(0.0f, 0.0f, 0.0f);
 
@@ -253,7 +251,7 @@ void Camera::update(Scene *scene)
 	ProjectionTransform screentocamera = projection_inverse(cameratoscreen);
 
 	rastertocamera = screentocamera * rastertoscreen;
-	full_rastertocamera = screentocamera * full_rastertoscreen;
+	ProjectionTransform full_rastertocamera = screentocamera * full_rastertoscreen;
 	cameratoraster = screentoraster * cameratoscreen;
 
 	cameratoworld = matrix;
@@ -629,7 +627,7 @@ float Camera::world_to_raster_size(float3 P)
 
 		if(offscreen_dicing_scale > 1.0f) {
 			float3 p = transform_point(&worldtocamera, P);
-			float3 v = transform_perspective(&full_rastertocamera, make_float3(full_width, full_height, 0.0f));
+			float3 v = transform_perspective(&rastertocamera, make_float3(width, height, 0.0f));
 
 			/* Create point clamped to frustum */
 			float3 c;
@@ -646,8 +644,8 @@ float Camera::world_to_raster_size(float3 P)
 	}
 	else if(type == CAMERA_PERSPECTIVE) {
 		/* Calculate as if point is directly ahead of the camera. */
-		float3 raster = make_float3(0.5f*full_width, 0.5f*full_height, 0.0f);
-		float3 Pcamera = transform_perspective(&full_rastertocamera, raster);
+		float3 raster = make_float3(0.5f*width, 0.5f*height, 0.0f);
+		float3 Pcamera = transform_perspective(&rastertocamera, raster);
 
 		/* dDdx */
 		float3 Ddiff = transform_direction(&cameratoworld, Pcamera);
@@ -730,21 +728,22 @@ float Camera::world_to_raster_size(float3 P)
 		 * point directly ahead seems to produce good enough results. */
 #if 0
 		float2 dir = direction_to_panorama(&kernel_camera, kernel_camera_motion.data(), normalize(D));
-		float3 raster = transform_perspective(&full_cameratoraster, make_float3(dir.x, dir.y, 0.0f));
+		float3 raster = transform_perspective(&cameratoraster, make_float3(dir.x, dir.y, 0.0f));
 
 		ray.t = 1.0f;
 		camera_sample_panorama(&kernel_camera, kernel_camera_motion.data(), raster.x, raster.y, 0.0f, 0.0f, &ray);
 		if(ray.t == 0.0f) {
 			/* No differentials, just use from directly ahead. */
-			camera_sample_panorama(&kernel_camera, kernel_camera_motion.data(), 0.5f*full_width, 0.5f*full_height, 0.0f, 0.0f, &ray);
+			camera_sample_panorama(&kernel_camera, kernel_camera_motion.data(), 0.5f*width, 0.5f*height, 0.0f, 0.0f, &ray);
 		}
 #else
-		camera_sample_panorama(&kernel_camera, kernel_camera_motion.data(), 0.5f*full_width, 0.5f*full_height, 0.0f, 0.0f, &ray);
+		camera_sample_panorama(&kernel_camera, kernel_camera_motion.data(), 0.5f*width, 0.5f*height, 0.0f, 0.0f, &ray);
 #endif
 
 		differential_transfer(&ray.dP, ray.dP, ray.D, ray.dD, ray.D, dist);
 
-		return max(len(ray.dP.dx),len(ray.dP.dy));
+		return max(len(ray.dP.dx) * (float(width)/float(full_width)),
+		           len(ray.dP.dy) * (float(height)/float(full_height)));
 	}
 
 	return res;

intern/cycles/render/camera.h

@@ -160,8 +160,6 @@ public:
 	ProjectionTransform rastertocamera;
 	ProjectionTransform cameratoraster;
 
-	ProjectionTransform full_rastertocamera;
-
 	float3 dx;
 	float3 dy;
 

intern/cycles/render/curves.h

@@ -75,7 +75,7 @@ public:
 	array<int> curve_firstkey;
 	array<int> curve_keynum;
 	array<float> curve_length;
-	array<float2> curve_uv;
+	array<float3> curve_uv;
 	array<float3> curve_vcol;
 
 	array<float3> curvekey_co;