c6275da852
Make volume stack allocated conditionally, potentially based on the actual nested level of objects in the scene. Currently the nested level is estimated by number of volume objects. This is a non-expensive check which is probably enough in practice to get almost perfect memory usage and performance. The conditional allocation is a bit tricky. For the CPU we declare and define maximum possible volume stack, because there are only that many integrator states on the CPU. On the GPU we declare outer SoA to have all volume stack elements, but only allocate actually needed ones. The actually used volume stack size is passed as a pre-processor, which seems to be easiest and fastest for the GPU state copy. There seems to be no speed regression in the demo files on RTX6000. Note that scenes with high nested level of volume will now be slower but correct. Differential Revision: https://developer.blender.org/D12759
282 lines
10 KiB
C
282 lines
10 KiB
C
/*
|
|
* Copyright 2011-2021 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.
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include "kernel/integrator/integrator_state.h"
|
|
#include "kernel/kernel_differential.h"
|
|
|
|
CCL_NAMESPACE_BEGIN
|
|
|
|
/* Ray */
|
|
|
|
ccl_device_forceinline void integrator_state_write_ray(INTEGRATOR_STATE_ARGS,
|
|
const Ray *ccl_restrict ray)
|
|
{
|
|
INTEGRATOR_STATE_WRITE(ray, P) = ray->P;
|
|
INTEGRATOR_STATE_WRITE(ray, D) = ray->D;
|
|
INTEGRATOR_STATE_WRITE(ray, t) = ray->t;
|
|
INTEGRATOR_STATE_WRITE(ray, time) = ray->time;
|
|
INTEGRATOR_STATE_WRITE(ray, dP) = ray->dP;
|
|
INTEGRATOR_STATE_WRITE(ray, dD) = ray->dD;
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_read_ray(INTEGRATOR_STATE_CONST_ARGS,
|
|
Ray *ccl_restrict ray)
|
|
{
|
|
ray->P = INTEGRATOR_STATE(ray, P);
|
|
ray->D = INTEGRATOR_STATE(ray, D);
|
|
ray->t = INTEGRATOR_STATE(ray, t);
|
|
ray->time = INTEGRATOR_STATE(ray, time);
|
|
ray->dP = INTEGRATOR_STATE(ray, dP);
|
|
ray->dD = INTEGRATOR_STATE(ray, dD);
|
|
}
|
|
|
|
/* Shadow Ray */
|
|
|
|
ccl_device_forceinline void integrator_state_write_shadow_ray(INTEGRATOR_STATE_ARGS,
|
|
const Ray *ccl_restrict ray)
|
|
{
|
|
INTEGRATOR_STATE_WRITE(shadow_ray, P) = ray->P;
|
|
INTEGRATOR_STATE_WRITE(shadow_ray, D) = ray->D;
|
|
INTEGRATOR_STATE_WRITE(shadow_ray, t) = ray->t;
|
|
INTEGRATOR_STATE_WRITE(shadow_ray, time) = ray->time;
|
|
INTEGRATOR_STATE_WRITE(shadow_ray, dP) = ray->dP;
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_read_shadow_ray(INTEGRATOR_STATE_CONST_ARGS,
|
|
Ray *ccl_restrict ray)
|
|
{
|
|
ray->P = INTEGRATOR_STATE(shadow_ray, P);
|
|
ray->D = INTEGRATOR_STATE(shadow_ray, D);
|
|
ray->t = INTEGRATOR_STATE(shadow_ray, t);
|
|
ray->time = INTEGRATOR_STATE(shadow_ray, time);
|
|
ray->dP = INTEGRATOR_STATE(shadow_ray, dP);
|
|
ray->dD = differential_zero_compact();
|
|
}
|
|
|
|
/* Intersection */
|
|
|
|
ccl_device_forceinline void integrator_state_write_isect(INTEGRATOR_STATE_ARGS,
|
|
const Intersection *ccl_restrict isect)
|
|
{
|
|
INTEGRATOR_STATE_WRITE(isect, t) = isect->t;
|
|
INTEGRATOR_STATE_WRITE(isect, u) = isect->u;
|
|
INTEGRATOR_STATE_WRITE(isect, v) = isect->v;
|
|
INTEGRATOR_STATE_WRITE(isect, object) = isect->object;
|
|
INTEGRATOR_STATE_WRITE(isect, prim) = isect->prim;
|
|
INTEGRATOR_STATE_WRITE(isect, type) = isect->type;
|
|
#ifdef __EMBREE__
|
|
INTEGRATOR_STATE_WRITE(isect, Ng) = isect->Ng;
|
|
#endif
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_read_isect(INTEGRATOR_STATE_CONST_ARGS,
|
|
Intersection *ccl_restrict isect)
|
|
{
|
|
isect->prim = INTEGRATOR_STATE(isect, prim);
|
|
isect->object = INTEGRATOR_STATE(isect, object);
|
|
isect->type = INTEGRATOR_STATE(isect, type);
|
|
isect->u = INTEGRATOR_STATE(isect, u);
|
|
isect->v = INTEGRATOR_STATE(isect, v);
|
|
isect->t = INTEGRATOR_STATE(isect, t);
|
|
#ifdef __EMBREE__
|
|
isect->Ng = INTEGRATOR_STATE(isect, Ng);
|
|
#endif
|
|
}
|
|
|
|
ccl_device_forceinline VolumeStack integrator_state_read_volume_stack(INTEGRATOR_STATE_CONST_ARGS,
|
|
int i)
|
|
{
|
|
VolumeStack entry = {INTEGRATOR_STATE_ARRAY(volume_stack, i, object),
|
|
INTEGRATOR_STATE_ARRAY(volume_stack, i, shader)};
|
|
return entry;
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_write_volume_stack(INTEGRATOR_STATE_ARGS,
|
|
int i,
|
|
VolumeStack entry)
|
|
{
|
|
INTEGRATOR_STATE_ARRAY_WRITE(volume_stack, i, object) = entry.object;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(volume_stack, i, shader) = entry.shader;
|
|
}
|
|
|
|
ccl_device_forceinline bool integrator_state_volume_stack_is_empty(INTEGRATOR_STATE_CONST_ARGS)
|
|
{
|
|
return (kernel_data.kernel_features & KERNEL_FEATURE_VOLUME) ?
|
|
INTEGRATOR_STATE_ARRAY(volume_stack, 0, shader) == SHADER_NONE :
|
|
true;
|
|
}
|
|
|
|
/* Shadow Intersection */
|
|
|
|
ccl_device_forceinline void integrator_state_write_shadow_isect(
|
|
INTEGRATOR_STATE_ARGS, const Intersection *ccl_restrict isect, const int index)
|
|
{
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, t) = isect->t;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, u) = isect->u;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, v) = isect->v;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, object) = isect->object;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, prim) = isect->prim;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, type) = isect->type;
|
|
#ifdef __EMBREE__
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_isect, index, Ng) = isect->Ng;
|
|
#endif
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_read_shadow_isect(INTEGRATOR_STATE_CONST_ARGS,
|
|
Intersection *ccl_restrict isect,
|
|
const int index)
|
|
{
|
|
isect->prim = INTEGRATOR_STATE_ARRAY(shadow_isect, index, prim);
|
|
isect->object = INTEGRATOR_STATE_ARRAY(shadow_isect, index, object);
|
|
isect->type = INTEGRATOR_STATE_ARRAY(shadow_isect, index, type);
|
|
isect->u = INTEGRATOR_STATE_ARRAY(shadow_isect, index, u);
|
|
isect->v = INTEGRATOR_STATE_ARRAY(shadow_isect, index, v);
|
|
isect->t = INTEGRATOR_STATE_ARRAY(shadow_isect, index, t);
|
|
#ifdef __EMBREE__
|
|
isect->Ng = INTEGRATOR_STATE_ARRAY(shadow_isect, index, Ng);
|
|
#endif
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_copy_volume_stack_to_shadow(INTEGRATOR_STATE_ARGS)
|
|
{
|
|
if (kernel_data.kernel_features & KERNEL_FEATURE_VOLUME) {
|
|
for (int i = 0; i < kernel_data.volume_stack_size; i++) {
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_volume_stack, i, object) = INTEGRATOR_STATE_ARRAY(
|
|
volume_stack, i, object);
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_volume_stack, i, shader) = INTEGRATOR_STATE_ARRAY(
|
|
volume_stack, i, shader);
|
|
}
|
|
}
|
|
}
|
|
|
|
ccl_device_forceinline VolumeStack
|
|
integrator_state_read_shadow_volume_stack(INTEGRATOR_STATE_CONST_ARGS, int i)
|
|
{
|
|
VolumeStack entry = {INTEGRATOR_STATE_ARRAY(shadow_volume_stack, i, object),
|
|
INTEGRATOR_STATE_ARRAY(shadow_volume_stack, i, shader)};
|
|
return entry;
|
|
}
|
|
|
|
ccl_device_forceinline bool integrator_state_shadow_volume_stack_is_empty(
|
|
INTEGRATOR_STATE_CONST_ARGS)
|
|
{
|
|
return (kernel_data.kernel_features & KERNEL_FEATURE_VOLUME) ?
|
|
INTEGRATOR_STATE_ARRAY(shadow_volume_stack, 0, shader) == SHADER_NONE :
|
|
true;
|
|
}
|
|
|
|
ccl_device_forceinline void integrator_state_write_shadow_volume_stack(INTEGRATOR_STATE_ARGS,
|
|
int i,
|
|
VolumeStack entry)
|
|
{
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_volume_stack, i, object) = entry.object;
|
|
INTEGRATOR_STATE_ARRAY_WRITE(shadow_volume_stack, i, shader) = entry.shader;
|
|
}
|
|
|
|
#if defined(__KERNEL_GPU__)
|
|
ccl_device_inline void integrator_state_copy_only(const IntegratorState to_state,
|
|
const IntegratorState state)
|
|
{
|
|
int index;
|
|
|
|
/* Rely on the compiler to optimize out unused assignments and `while(false)`'s. */
|
|
|
|
# define KERNEL_STRUCT_BEGIN(name) \
|
|
index = 0; \
|
|
do {
|
|
|
|
# define KERNEL_STRUCT_MEMBER(parent_struct, type, name, feature) \
|
|
if (kernel_integrator_state.parent_struct.name != nullptr) { \
|
|
kernel_integrator_state.parent_struct.name[to_state] = \
|
|
kernel_integrator_state.parent_struct.name[state]; \
|
|
}
|
|
|
|
# define KERNEL_STRUCT_ARRAY_MEMBER(parent_struct, type, name, feature) \
|
|
if (kernel_integrator_state.parent_struct[index].name != nullptr) { \
|
|
kernel_integrator_state.parent_struct[index].name[to_state] = \
|
|
kernel_integrator_state.parent_struct[index].name[state]; \
|
|
}
|
|
|
|
# define KERNEL_STRUCT_END(name) \
|
|
} \
|
|
while (false) \
|
|
;
|
|
|
|
# define KERNEL_STRUCT_END_ARRAY(name, cpu_array_size, gpu_array_size) \
|
|
++index; \
|
|
} \
|
|
while (index < gpu_array_size) \
|
|
;
|
|
|
|
# define KERNEL_STRUCT_VOLUME_STACK_SIZE kernel_data.volume_stack_size
|
|
|
|
# include "kernel/integrator/integrator_state_template.h"
|
|
|
|
# undef KERNEL_STRUCT_BEGIN
|
|
# undef KERNEL_STRUCT_MEMBER
|
|
# undef KERNEL_STRUCT_ARRAY_MEMBER
|
|
# undef KERNEL_STRUCT_END
|
|
# undef KERNEL_STRUCT_END_ARRAY
|
|
# undef KERNEL_STRUCT_VOLUME_STACK_SIZE
|
|
}
|
|
|
|
ccl_device_inline void integrator_state_move(const IntegratorState to_state,
|
|
const IntegratorState state)
|
|
{
|
|
integrator_state_copy_only(to_state, state);
|
|
|
|
INTEGRATOR_STATE_WRITE(path, queued_kernel) = 0;
|
|
INTEGRATOR_STATE_WRITE(shadow_path, queued_kernel) = 0;
|
|
}
|
|
|
|
#endif
|
|
|
|
/* NOTE: Leaves kernel scheduling information untouched. Use INIT semantic for one of the paths
|
|
* after this function. */
|
|
ccl_device_inline void integrator_state_shadow_catcher_split(INTEGRATOR_STATE_ARGS)
|
|
{
|
|
#if defined(__KERNEL_GPU__)
|
|
const IntegratorState to_state = atomic_fetch_and_add_uint32(
|
|
&kernel_integrator_state.next_shadow_catcher_path_index[0], 1);
|
|
|
|
integrator_state_copy_only(to_state, state);
|
|
|
|
kernel_integrator_state.path.flag[to_state] |= PATH_RAY_SHADOW_CATCHER_PASS;
|
|
|
|
/* Sanity check: expect to split in the intersect-closest kernel, where there is no shadow ray
|
|
* and no sorting yet. */
|
|
kernel_assert(INTEGRATOR_STATE(shadow_path, queued_kernel) == 0);
|
|
kernel_assert(kernel_integrator_state.sort_key_counter[INTEGRATOR_STATE(path, queued_kernel)] ==
|
|
nullptr);
|
|
#else
|
|
|
|
IntegratorStateCPU *ccl_restrict split_state = state + 1;
|
|
|
|
/* Only copy the required subset, since shadow intersections are big and irrelevant here. */
|
|
split_state->path = state->path;
|
|
split_state->ray = state->ray;
|
|
split_state->isect = state->isect;
|
|
memcpy(split_state->volume_stack, state->volume_stack, sizeof(state->volume_stack));
|
|
split_state->shadow_path = state->shadow_path;
|
|
|
|
split_state->path.flag |= PATH_RAY_SHADOW_CATCHER_PASS;
|
|
#endif
|
|
}
|
|
|
|
CCL_NAMESPACE_END
|