
With the compute pipeline calculation can be offloaded to the GPU. This patch only adds the framework for compute. So no changes for users at this moment. NOTE: As this is an OpenGL4.3 feature it must always have a fallback. Use `GPU_compute_shader_support` to check if compute pipeline can be used. Check `gpu_shader_compute*` test cases for usage. This patch also adds support for shader storage buffer objects and device only vertex/index buffers. An alternative that had been discussed was adding this to the `GPUBatch`, this was eventually not chosen as it would lead to more code when used as part of a shading group. The idea is that we add an `eDRWCommandType` in the near future. Reviewed By: fclem Differential Revision: https://developer.blender.org/D10913
469 lines
12 KiB
C++
469 lines
12 KiB
C++
/*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version 2
|
|
* of the License, or (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
*
|
|
* The Original Code is Copyright (C) 2020 Blender Foundation.
|
|
* All rights reserved.
|
|
*/
|
|
|
|
/** \file
|
|
* \ingroup gpu
|
|
*/
|
|
|
|
#include "BKE_global.h"
|
|
|
|
#include "BLI_string.h"
|
|
#include "BLI_vector.hh"
|
|
|
|
#include "GPU_capabilities.h"
|
|
#include "GPU_platform.h"
|
|
|
|
#include "gl_backend.hh"
|
|
#include "gl_debug.hh"
|
|
#include "gl_vertex_buffer.hh"
|
|
|
|
#include "gl_shader.hh"
|
|
#include "gl_shader_interface.hh"
|
|
|
|
using namespace blender;
|
|
using namespace blender::gpu;
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Creation / Destruction
|
|
* \{ */
|
|
|
|
GLShader::GLShader(const char *name) : Shader(name)
|
|
{
|
|
#if 0 /* Would be nice to have, but for now the Deferred compilation \
|
|
* does not have a GPUContext. */
|
|
BLI_assert(GLContext::get() != NULL);
|
|
#endif
|
|
shader_program_ = glCreateProgram();
|
|
|
|
debug::object_label(GL_PROGRAM, shader_program_, name);
|
|
}
|
|
|
|
GLShader::~GLShader()
|
|
{
|
|
#if 0 /* Would be nice to have, but for now the Deferred compilation \
|
|
* does not have a GPUContext. */
|
|
BLI_assert(GLContext::get() != NULL);
|
|
#endif
|
|
/* Invalid handles are silently ignored. */
|
|
glDeleteShader(vert_shader_);
|
|
glDeleteShader(geom_shader_);
|
|
glDeleteShader(frag_shader_);
|
|
glDeleteShader(compute_shader_);
|
|
glDeleteProgram(shader_program_);
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Shader stage creation
|
|
* \{ */
|
|
|
|
static char *glsl_patch_default_get()
|
|
{
|
|
/** Used for shader patching. Init once. */
|
|
static char patch[512] = "\0";
|
|
if (patch[0] != '\0') {
|
|
return patch;
|
|
}
|
|
|
|
size_t slen = 0;
|
|
/* Version need to go first. */
|
|
STR_CONCAT(patch, slen, "#version 330\n");
|
|
|
|
/* Enable extensions for features that are not part of our base GLSL version
|
|
* don't use an extension for something already available! */
|
|
if (GLContext::texture_gather_support) {
|
|
STR_CONCAT(patch, slen, "#extension GL_ARB_texture_gather: enable\n");
|
|
/* Some drivers don't agree on GLEW_ARB_texture_gather and the actual support in the
|
|
* shader so double check the preprocessor define (see T56544). */
|
|
STR_CONCAT(patch, slen, "#ifdef GL_ARB_texture_gather\n");
|
|
STR_CONCAT(patch, slen, "# define GPU_ARB_texture_gather\n");
|
|
STR_CONCAT(patch, slen, "#endif\n");
|
|
}
|
|
if (GLContext::shader_draw_parameters_support) {
|
|
STR_CONCAT(patch, slen, "#extension GL_ARB_shader_draw_parameters : enable\n");
|
|
STR_CONCAT(patch, slen, "#define GPU_ARB_shader_draw_parameters\n");
|
|
}
|
|
if (GLContext::texture_cube_map_array_support) {
|
|
STR_CONCAT(patch, slen, "#extension GL_ARB_texture_cube_map_array : enable\n");
|
|
STR_CONCAT(patch, slen, "#define GPU_ARB_texture_cube_map_array\n");
|
|
}
|
|
|
|
/* Derivative sign can change depending on implementation. */
|
|
STR_CONCATF(patch, slen, "#define DFDX_SIGN %1.1f\n", GLContext::derivative_signs[0]);
|
|
STR_CONCATF(patch, slen, "#define DFDY_SIGN %1.1f\n", GLContext::derivative_signs[1]);
|
|
|
|
BLI_assert(slen < sizeof(patch));
|
|
return patch;
|
|
}
|
|
|
|
static char *glsl_patch_compute_get()
|
|
{
|
|
/** Used for shader patching. Init once. */
|
|
static char patch[512] = "\0";
|
|
if (patch[0] != '\0') {
|
|
return patch;
|
|
}
|
|
|
|
size_t slen = 0;
|
|
/* Version need to go first. */
|
|
STR_CONCAT(patch, slen, "#version 430\n");
|
|
STR_CONCAT(patch, slen, "#extension GL_ARB_compute_shader :enable\n");
|
|
BLI_assert(slen < sizeof(patch));
|
|
return patch;
|
|
}
|
|
|
|
char *GLShader::glsl_patch_get(GLenum gl_stage)
|
|
{
|
|
if (gl_stage == GL_COMPUTE_SHADER) {
|
|
return glsl_patch_compute_get();
|
|
}
|
|
return glsl_patch_default_get();
|
|
}
|
|
|
|
/* Create, compile and attach the shader stage to the shader program. */
|
|
GLuint GLShader::create_shader_stage(GLenum gl_stage, MutableSpan<const char *> sources)
|
|
{
|
|
GLuint shader = glCreateShader(gl_stage);
|
|
if (shader == 0) {
|
|
fprintf(stderr, "GLShader: Error: Could not create shader object.");
|
|
return 0;
|
|
}
|
|
|
|
/* Patch the shader code using the first source slot. */
|
|
sources[0] = glsl_patch_get(gl_stage);
|
|
|
|
glShaderSource(shader, sources.size(), sources.data(), nullptr);
|
|
glCompileShader(shader);
|
|
|
|
GLint status;
|
|
glGetShaderiv(shader, GL_COMPILE_STATUS, &status);
|
|
if (!status || (G.debug & G_DEBUG_GPU)) {
|
|
char log[5000] = "";
|
|
glGetShaderInfoLog(shader, sizeof(log), nullptr, log);
|
|
if (log[0] != '\0') {
|
|
switch (gl_stage) {
|
|
case GL_VERTEX_SHADER:
|
|
this->print_log(sources, log, "VertShader", !status);
|
|
break;
|
|
case GL_GEOMETRY_SHADER:
|
|
this->print_log(sources, log, "GeomShader", !status);
|
|
break;
|
|
case GL_FRAGMENT_SHADER:
|
|
this->print_log(sources, log, "FragShader", !status);
|
|
break;
|
|
case GL_COMPUTE_SHADER:
|
|
this->print_log(sources, log, "ComputeShader", !status);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
if (!status) {
|
|
glDeleteShader(shader);
|
|
compilation_failed_ = true;
|
|
return 0;
|
|
}
|
|
|
|
debug::object_label(gl_stage, shader, name);
|
|
|
|
glAttachShader(shader_program_, shader);
|
|
return shader;
|
|
}
|
|
|
|
void GLShader::vertex_shader_from_glsl(MutableSpan<const char *> sources)
|
|
{
|
|
vert_shader_ = this->create_shader_stage(GL_VERTEX_SHADER, sources);
|
|
}
|
|
|
|
void GLShader::geometry_shader_from_glsl(MutableSpan<const char *> sources)
|
|
{
|
|
geom_shader_ = this->create_shader_stage(GL_GEOMETRY_SHADER, sources);
|
|
}
|
|
|
|
void GLShader::fragment_shader_from_glsl(MutableSpan<const char *> sources)
|
|
{
|
|
frag_shader_ = this->create_shader_stage(GL_FRAGMENT_SHADER, sources);
|
|
}
|
|
|
|
void GLShader::compute_shader_from_glsl(MutableSpan<const char *> sources)
|
|
{
|
|
compute_shader_ = this->create_shader_stage(GL_COMPUTE_SHADER, sources);
|
|
}
|
|
|
|
bool GLShader::finalize()
|
|
{
|
|
if (compilation_failed_) {
|
|
return false;
|
|
}
|
|
|
|
glLinkProgram(shader_program_);
|
|
|
|
GLint status;
|
|
glGetProgramiv(shader_program_, GL_LINK_STATUS, &status);
|
|
if (!status) {
|
|
char log[5000];
|
|
glGetProgramInfoLog(shader_program_, sizeof(log), nullptr, log);
|
|
Span<const char *> sources;
|
|
this->print_log(sources, log, "Linking", true);
|
|
return false;
|
|
}
|
|
|
|
interface = new GLShaderInterface(shader_program_);
|
|
|
|
return true;
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Binding
|
|
* \{ */
|
|
|
|
void GLShader::bind()
|
|
{
|
|
BLI_assert(shader_program_ != 0);
|
|
glUseProgram(shader_program_);
|
|
}
|
|
|
|
void GLShader::unbind()
|
|
{
|
|
#ifndef NDEBUG
|
|
glUseProgram(0);
|
|
#endif
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Transform feedback
|
|
*
|
|
* TODO(fclem): Should be replaced by compute shaders.
|
|
* \{ */
|
|
|
|
/* Should be called before linking. */
|
|
void GLShader::transform_feedback_names_set(Span<const char *> name_list,
|
|
const eGPUShaderTFBType geom_type)
|
|
{
|
|
glTransformFeedbackVaryings(
|
|
shader_program_, name_list.size(), name_list.data(), GL_INTERLEAVED_ATTRIBS);
|
|
transform_feedback_type_ = geom_type;
|
|
}
|
|
|
|
bool GLShader::transform_feedback_enable(GPUVertBuf *buf_)
|
|
{
|
|
if (transform_feedback_type_ == GPU_SHADER_TFB_NONE) {
|
|
return false;
|
|
}
|
|
|
|
GLVertBuf *buf = static_cast<GLVertBuf *>(unwrap(buf_));
|
|
|
|
BLI_assert(buf->vbo_id_ != 0);
|
|
|
|
glBindBufferBase(GL_TRANSFORM_FEEDBACK_BUFFER, 0, buf->vbo_id_);
|
|
|
|
switch (transform_feedback_type_) {
|
|
case GPU_SHADER_TFB_POINTS:
|
|
glBeginTransformFeedback(GL_POINTS);
|
|
break;
|
|
case GPU_SHADER_TFB_LINES:
|
|
glBeginTransformFeedback(GL_LINES);
|
|
break;
|
|
case GPU_SHADER_TFB_TRIANGLES:
|
|
glBeginTransformFeedback(GL_TRIANGLES);
|
|
break;
|
|
default:
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void GLShader::transform_feedback_disable()
|
|
{
|
|
glEndTransformFeedback();
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Uniforms setters
|
|
* \{ */
|
|
|
|
void GLShader::uniform_float(int location, int comp_len, int array_size, const float *data)
|
|
{
|
|
switch (comp_len) {
|
|
case 1:
|
|
glUniform1fv(location, array_size, data);
|
|
break;
|
|
case 2:
|
|
glUniform2fv(location, array_size, data);
|
|
break;
|
|
case 3:
|
|
glUniform3fv(location, array_size, data);
|
|
break;
|
|
case 4:
|
|
glUniform4fv(location, array_size, data);
|
|
break;
|
|
case 9:
|
|
glUniformMatrix3fv(location, array_size, 0, data);
|
|
break;
|
|
case 16:
|
|
glUniformMatrix4fv(location, array_size, 0, data);
|
|
break;
|
|
default:
|
|
BLI_assert(0);
|
|
break;
|
|
}
|
|
}
|
|
|
|
void GLShader::uniform_int(int location, int comp_len, int array_size, const int *data)
|
|
{
|
|
switch (comp_len) {
|
|
case 1:
|
|
glUniform1iv(location, array_size, data);
|
|
break;
|
|
case 2:
|
|
glUniform2iv(location, array_size, data);
|
|
break;
|
|
case 3:
|
|
glUniform3iv(location, array_size, data);
|
|
break;
|
|
case 4:
|
|
glUniform4iv(location, array_size, data);
|
|
break;
|
|
default:
|
|
BLI_assert(0);
|
|
break;
|
|
}
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name GPUVertFormat from Shader
|
|
* \{ */
|
|
|
|
static uint calc_component_size(const GLenum gl_type)
|
|
{
|
|
switch (gl_type) {
|
|
case GL_FLOAT_VEC2:
|
|
case GL_INT_VEC2:
|
|
case GL_UNSIGNED_INT_VEC2:
|
|
return 2;
|
|
case GL_FLOAT_VEC3:
|
|
case GL_INT_VEC3:
|
|
case GL_UNSIGNED_INT_VEC3:
|
|
return 3;
|
|
case GL_FLOAT_VEC4:
|
|
case GL_FLOAT_MAT2:
|
|
case GL_INT_VEC4:
|
|
case GL_UNSIGNED_INT_VEC4:
|
|
return 4;
|
|
case GL_FLOAT_MAT3:
|
|
return 9;
|
|
case GL_FLOAT_MAT4:
|
|
return 16;
|
|
case GL_FLOAT_MAT2x3:
|
|
case GL_FLOAT_MAT3x2:
|
|
return 6;
|
|
case GL_FLOAT_MAT2x4:
|
|
case GL_FLOAT_MAT4x2:
|
|
return 8;
|
|
case GL_FLOAT_MAT3x4:
|
|
case GL_FLOAT_MAT4x3:
|
|
return 12;
|
|
default:
|
|
return 1;
|
|
}
|
|
}
|
|
|
|
static void get_fetch_mode_and_comp_type(int gl_type,
|
|
GPUVertCompType *r_comp_type,
|
|
GPUVertFetchMode *r_fetch_mode)
|
|
{
|
|
switch (gl_type) {
|
|
case GL_FLOAT:
|
|
case GL_FLOAT_VEC2:
|
|
case GL_FLOAT_VEC3:
|
|
case GL_FLOAT_VEC4:
|
|
case GL_FLOAT_MAT2:
|
|
case GL_FLOAT_MAT3:
|
|
case GL_FLOAT_MAT4:
|
|
case GL_FLOAT_MAT2x3:
|
|
case GL_FLOAT_MAT2x4:
|
|
case GL_FLOAT_MAT3x2:
|
|
case GL_FLOAT_MAT3x4:
|
|
case GL_FLOAT_MAT4x2:
|
|
case GL_FLOAT_MAT4x3:
|
|
*r_comp_type = GPU_COMP_F32;
|
|
*r_fetch_mode = GPU_FETCH_FLOAT;
|
|
break;
|
|
case GL_INT:
|
|
case GL_INT_VEC2:
|
|
case GL_INT_VEC3:
|
|
case GL_INT_VEC4:
|
|
*r_comp_type = GPU_COMP_I32;
|
|
*r_fetch_mode = GPU_FETCH_INT;
|
|
break;
|
|
case GL_UNSIGNED_INT:
|
|
case GL_UNSIGNED_INT_VEC2:
|
|
case GL_UNSIGNED_INT_VEC3:
|
|
case GL_UNSIGNED_INT_VEC4:
|
|
*r_comp_type = GPU_COMP_U32;
|
|
*r_fetch_mode = GPU_FETCH_INT;
|
|
break;
|
|
default:
|
|
BLI_assert(0);
|
|
}
|
|
}
|
|
|
|
void GLShader::vertformat_from_shader(GPUVertFormat *format) const
|
|
{
|
|
GPU_vertformat_clear(format);
|
|
|
|
GLint attr_len;
|
|
glGetProgramiv(shader_program_, GL_ACTIVE_ATTRIBUTES, &attr_len);
|
|
|
|
for (int i = 0; i < attr_len; i++) {
|
|
char name[256];
|
|
GLenum gl_type;
|
|
GLint size;
|
|
glGetActiveAttrib(shader_program_, i, sizeof(name), nullptr, &size, &gl_type, name);
|
|
|
|
/* Ignore OpenGL names like `gl_BaseInstanceARB`, `gl_InstanceID` and `gl_VertexID`. */
|
|
if (glGetAttribLocation(shader_program_, name) == -1) {
|
|
continue;
|
|
}
|
|
|
|
GPUVertCompType comp_type;
|
|
GPUVertFetchMode fetch_mode;
|
|
get_fetch_mode_and_comp_type(gl_type, &comp_type, &fetch_mode);
|
|
|
|
int comp_len = calc_component_size(gl_type) * size;
|
|
|
|
GPU_vertformat_attr_add(format, name, comp_type, comp_len, fetch_mode);
|
|
}
|
|
}
|
|
|
|
int GLShader::program_handle_get() const
|
|
{
|
|
return (int)this->shader_program_;
|
|
}
|
|
|
|
/** \} */
|