8dce2a422b
On lower end hardware the film accumulation has bad performance. Sometimes upto 10ms. This PR improves the performance somewhat by adding a specialization constant around the renderpasses that are actually needed for rendering, the number of samples and if reprojection is enabled. `enabled_categories`: Based on the enabled render passes some outer loops are enabled/disabled that handle the specific render passes. This improves the performance as no memory will be reserved for branches that are never accessed. `samples_len` & `use_reprojection`: GPU compilers tend to optimize texture fetches when they to the outer loop. This is only possible when the inner loop can be unrolled. In the case of the film accumulation the inner loop couldn't be unrolled. By adding a specialization constant would allow unrolling of the inner loop. On old or low-end devices the improvement is around 40%. On newer devices the improvement is 50+%. Performance of this shader is similar to the godot. | GPU | Before | New | |----------------------|--------|-------| | NVIDIA GTX 760 | 3.5ms | 2.4ms | | GFX1036 (RDNA2 iGPU) | 9.9ms | 6.2ms | | AMD Radeon Pro W7500 | 2.1ms | 0.9ms | Pull Request: https://projects.blender.org/blender/blender/pulls/118385
117 lines
3.4 KiB
C
117 lines
3.4 KiB
C
/* SPDX-FileCopyrightText: 2022 Blender Authors
|
|
*
|
|
* SPDX-License-Identifier: GPL-2.0-or-later */
|
|
|
|
/** \file
|
|
* \ingroup gpu
|
|
*
|
|
* Glue definition to make shared declaration of struct & functions work in both C / C++ and GLSL.
|
|
* We use the same vector and matrix types as Blender C++. Some math functions are defined to use
|
|
* the float version to match the GLSL syntax.
|
|
* This file can be used for C & C++ code and the syntax used should follow the same rules.
|
|
* Some preprocessing is done by the GPU back-end to make it GLSL compatible.
|
|
*
|
|
* IMPORTANT:
|
|
* - Always use `u` suffix for enum values. GLSL do not support implicit cast.
|
|
* - Define all values. This is in order to simplify custom pre-processor code.
|
|
* - (C++ only) Always use `uint32_t` as underlying type (`enum eMyEnum : uint32_t`).
|
|
* - (C only) do NOT use the enum type inside UBO/SSBO structs and use `uint` instead.
|
|
* - Use float suffix by default for float literals to avoid double promotion in C++.
|
|
* - Pack one float or int after a vec3/ivec3 to fulfill alignment rules.
|
|
*
|
|
* NOTE: Due to alignment restriction and buggy drivers, do not try to use mat3 inside structs.
|
|
* NOTE: (UBO only) Do not use arrays of float. They are padded to arrays of vec4 and are not worth
|
|
* it. This does not apply to SSBO.
|
|
*
|
|
* IMPORTANT: Do not forget to align mat4, vec3 and vec4 to 16 bytes, and vec2 to 8 bytes.
|
|
*
|
|
* NOTE: You can use bool type using bool1 a int boolean type matching the GLSL type.
|
|
*/
|
|
|
|
#ifdef GPU_SHADER
|
|
# define BLI_STATIC_ASSERT(cond, msg)
|
|
# define BLI_STATIC_ASSERT_ALIGN(type_, align_)
|
|
# define BLI_STATIC_ASSERT_SIZE(type_, size_)
|
|
# define ENUM_OPERATORS(a, b)
|
|
# define static
|
|
# define inline
|
|
# define cosf cos
|
|
# define sinf sin
|
|
# define tanf tan
|
|
# define acosf acos
|
|
# define asinf asin
|
|
# define atanf atan
|
|
# define floorf floor
|
|
# define ceilf ceil
|
|
# define sqrtf sqrt
|
|
# define expf exp
|
|
|
|
# define bool1 bool
|
|
/* Type name collision with Metal shading language - These type-names are already defined. */
|
|
# ifndef GPU_METAL
|
|
# define float2 vec2
|
|
# define float3 vec3
|
|
# define float3x4 mat3x4
|
|
# define float4 vec4
|
|
# define float4x4 mat4
|
|
# define int2 ivec2
|
|
# define int3 ivec3
|
|
# define int4 ivec4
|
|
# define uint2 uvec2
|
|
# define uint3 uvec3
|
|
# define uint4 uvec4
|
|
# define bool2 bvec2
|
|
# define bool3 bvec3
|
|
# define bool4 bvec4
|
|
# define packed_float3 vec3
|
|
# define packed_int3 int3
|
|
# endif
|
|
|
|
#else /* C / C++ */
|
|
# pragma once
|
|
|
|
# include "BLI_assert.h"
|
|
|
|
# ifdef __cplusplus
|
|
# include "BLI_math_matrix_types.hh"
|
|
# include "BLI_math_vector_types.hh"
|
|
using blender::float2;
|
|
using blender::float3;
|
|
using blender::float3x4;
|
|
using blender::float4;
|
|
using blender::float4x4;
|
|
using blender::int2;
|
|
using blender::int3;
|
|
using blender::int4;
|
|
using blender::uint2;
|
|
using blender::uint3;
|
|
using blender::uint4;
|
|
using bool1 = int;
|
|
using bool2 = blender::int2;
|
|
using bool3 = blender::int3;
|
|
using bool4 = blender::int4;
|
|
using packed_float3 = blender::float3;
|
|
using packed_int3 = blender::int3;
|
|
|
|
# else /* C */
|
|
typedef float float2[2];
|
|
typedef float float3[3];
|
|
typedef float float4[4];
|
|
typedef float float4x4[4][4];
|
|
typedef float float3x4[3][4];
|
|
typedef int int2[2];
|
|
typedef int int3[2];
|
|
typedef int int4[4];
|
|
typedef uint uint2[2];
|
|
typedef uint uint3[3];
|
|
typedef uint uint4[4];
|
|
typedef int bool1;
|
|
typedef int bool2[2];
|
|
typedef int bool3[2];
|
|
typedef int bool4[4];
|
|
typedef float3 packed_float3;
|
|
typedef int3 packed_int3;
|
|
# endif
|
|
|
|
#endif
|