Cr8-xyz/blender
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Code refactor: remove rng_state buffer and compute hash on the fly.

Browse Source 
A little faster on some benchmark scenes, a little slower on others, seems
about performance neutral on average and saves a little memory.
This commit is contained in:
Brecht Van Lommel
2017-09-27 00:39:53 +02:00
parent 5b7d6ea54b
commit e3e16cecc4
25 changed files with 21 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
intern/cycles/device/device_cpu.cpp
View File

@@ -171,7 +171,7 @@ public:
DeviceRequestedFeatures requested_features;
KernelFunctions<void(*)(KernelGlobals *, float *, unsigned int *, int, int, int, int, int)> path_trace_kernel;
KernelFunctions<void(*)(KernelGlobals *, float *, int, int, int, int, int)> path_trace_kernel;
KernelFunctions<void(*)(KernelGlobals *, uchar4 *, float *, float, int, int, int, int)> convert_to_half_float_kernel;
KernelFunctions<void(*)(KernelGlobals *, uchar4 *, float *, float, int, int, int, int)> convert_to_byte_kernel;
KernelFunctions<void(*)(KernelGlobals *, uint4 *, float4 *, float*, int, int, int, int, int)> shader_kernel;
@@ -192,7 +192,7 @@ public:
KernelFunctions<void(*)(int, int, int, int, int, float*, int*, float*, float3*, int*, int)> filter_finalize_kernel;
KernelFunctions<void(*)(KernelGlobals *, ccl_constant KernelData*, ccl_global void*, int, ccl_global char*,
ccl_global uint*, int, int, int, int, int, int, int, int, ccl_global int*, int,
int, int, int, int, int, int, int, int, ccl_global int*, int,
ccl_global char*, ccl_global unsigned int*, unsigned int, ccl_global float*)> data_init_kernel;
unordered_map<string, KernelFunctions<void(*)(KernelGlobals*, KernelData*)> > split_kernels;
@@ -617,7 +617,6 @@ public:
void path_trace(DeviceTask &task, RenderTile &tile, KernelGlobals *kg)
{
float *render_buffer = (float*)tile.buffer;
uint *rng_state = (uint*)tile.rng_state;
int start_sample = tile.start_sample;
int end_sample = tile.start_sample + tile.num_samples;
@@ -629,7 +628,7 @@ public:
for(int y = tile.y; y < tile.y + tile.h; y++) {
for(int x = tile.x; x < tile.x + tile.w; x++) {
path_trace_kernel()(kg, render_buffer, rng_state,
path_trace_kernel()(kg, render_buffer,
sample, x, y, tile.offset, tile.stride);
}
}
@@ -913,7 +912,6 @@ bool CPUSplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim,
(void*)split_data.device_pointer,
num_global_elements,
(char*)ray_state.device_pointer,
(uint*)rtile.rng_state,
rtile.start_sample,
rtile.start_sample + rtile.num_samples,
rtile.x,

4
intern/cycles/device/device_cuda.cpp
View File

@@ -1322,7 +1322,6 @@ public:
wtile->start_sample = sample;
wtile->num_samples = 1;
wtile->buffer = (float*)cuda_device_ptr(rtile.buffer);
wtile->rng_state = (uint*)cuda_device_ptr(rtile.rng_state);
mem_alloc("work_tiles", work_tiles, MEM_READ_ONLY);
mem_copy_to(work_tiles);
@@ -1945,7 +1944,6 @@ bool CUDASplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim
CUdeviceptr d_use_queues_flag = device->cuda_device_ptr(use_queues_flag.device_pointer);
CUdeviceptr d_work_pool_wgs = device->cuda_device_ptr(work_pool_wgs.device_pointer);
CUdeviceptr d_rng_state = device->cuda_device_ptr(rtile.rng_state);
CUdeviceptr d_buffer = device->cuda_device_ptr(rtile.buffer);
int end_sample = rtile.start_sample + rtile.num_samples;
@@ -1955,7 +1953,6 @@ bool CUDASplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim
CUdeviceptr* split_data_buffer;
int* num_elements;
CUdeviceptr* ray_state;
CUdeviceptr* rng_state;
int* start_sample;
int* end_sample;
int* sx;
@@ -1976,7 +1973,6 @@ bool CUDASplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim
&d_split_data,
&num_global_elements,
&d_ray_state,
&d_rng_state,
&rtile.start_sample,
&end_sample,
&rtile.x,

1
intern/cycles/device/device_multi.cpp
View File

@@ -281,7 +281,6 @@ public:
foreach(SubDevice& sub, devices) {
if(sub.device == sub_device) {
if(tile.buffer) tile.buffer = sub.ptr_map[tile.buffer];
if(tile.rng_state) tile.rng_state = sub.ptr_map[tile.rng_state];
}
}
}

2
intern/cycles/device/device_network.cpp
View File

@@ -737,7 +737,6 @@ protected:
tile = entry.tile;
if(tile.buffer) tile.buffer = ptr_map[tile.buffer];
if(tile.rng_state) tile.rng_state = ptr_map[tile.rng_state];
result = true;
break;
@@ -769,7 +768,6 @@ protected:
thread_scoped_lock acquire_lock(acquire_mutex);
if(tile.buffer) tile.buffer = ptr_imap[tile.buffer];
if(tile.rng_state) tile.rng_state = ptr_imap[tile.rng_state];
{
thread_scoped_lock lock(rpc_lock);

4
intern/cycles/device/device_network.h
View File

@@ -142,7 +142,7 @@ public:
archive & tile.x & tile.y & tile.w & tile.h;
archive & tile.start_sample & tile.num_samples & tile.sample;
archive & tile.resolution & tile.offset & tile.stride;
archive & tile.buffer & tile.rng_state;
archive & tile.buffer;
}
void write()
@@ -303,7 +303,7 @@ public:
*archive & tile.x & tile.y & tile.w & tile.h;
*archive & tile.start_sample & tile.num_samples & tile.sample;
*archive & tile.resolution & tile.offset & tile.stride;
*archive & tile.buffer & tile.rng_state;
*archive & tile.buffer;
tile.buffers = NULL;
}

4
intern/cycles/device/opencl/opencl_mega.cpp
View File

@@ -62,7 +62,6 @@ public:
/* Cast arguments to cl types. */
cl_mem d_data = CL_MEM_PTR(const_mem_map["__data"]->device_pointer);
cl_mem d_buffer = CL_MEM_PTR(rtile.buffer);
cl_mem d_rng_state = CL_MEM_PTR(rtile.rng_state);
cl_int d_x = rtile.x;
cl_int d_y = rtile.y;
cl_int d_w = rtile.w;
@@ -79,8 +78,7 @@ public:
kernel_set_args(ckPathTraceKernel,
0,
d_data,
d_buffer,
d_rng_state);
d_buffer);
set_kernel_arg_buffers(ckPathTraceKernel, &start_arg_index);

8
intern/cycles/device/opencl/opencl_split.cpp
View File

@@ -192,7 +192,6 @@ struct CachedSplitMemory {
int id;
device_memory *split_data;
device_memory *ray_state;
device_ptr *rng_state;
device_memory *queue_index;
device_memory *use_queues_flag;
device_memory *work_pools;
@@ -225,8 +224,7 @@ public:
kg,
data,
*cached_memory.split_data,
*cached_memory.ray_state,
*cached_memory.rng_state);
*cached_memory.ray_state);
device->set_kernel_arg_buffers(program(), &start_arg_index);
@@ -356,8 +354,7 @@ public:
kernel_data,
split_data,
num_global_elements,
ray_state,
rtile.rng_state);
ray_state);
device->set_kernel_arg_buffers(device->program_data_init(), &start_arg_index);
@@ -401,7 +398,6 @@ public:
cached_memory.split_data = &split_data;
cached_memory.ray_state = &ray_state;
cached_memory.rng_state = &rtile.rng_state;
cached_memory.queue_index = &queue_index;
cached_memory.use_queues_flag = &use_queues_flag;
cached_memory.work_pools = &work_pool_wgs;

5
intern/cycles/kernel/kernel_path.h
View File

@@ -672,21 +672,20 @@ ccl_device_forceinline void kernel_path_integrate(
}
ccl_device void kernel_path_trace(KernelGlobals *kg,
ccl_global float *buffer, ccl_global uint *rng_state,
ccl_global float *buffer,
int sample, int x, int y, int offset, int stride)
{
/* buffer offset */
int index = offset + x + y*stride;
int pass_stride = kernel_data.film.pass_stride;
rng_state += index;
buffer += index*pass_stride;
/* Initialize random numbers and sample ray. */
uint rng_hash;
Ray ray;
kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng_hash, &ray);
kernel_path_trace_setup(kg, sample, x, y, &rng_hash, &ray);
if(ray.t == 0.0f) {
kernel_write_result(kg, buffer, sample, NULL);

5
intern/cycles/kernel/kernel_path_branched.h
View File

@@ -538,21 +538,20 @@ ccl_device void kernel_branched_path_integrate(KernelGlobals *kg,
}
ccl_device void kernel_branched_path_trace(KernelGlobals *kg,
ccl_global float *buffer, ccl_global uint *rng_state,
ccl_global float *buffer,
int sample, int x, int y, int offset, int stride)
{
/* buffer offset */
int index = offset + x + y*stride;
int pass_stride = kernel_data.film.pass_stride;
rng_state += index;
buffer += index*pass_stride;
/* initialize random numbers and ray */
uint rng_hash;
Ray ray;
kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng_hash, &ray);
kernel_path_trace_setup(kg, sample, x, y, &rng_hash, &ray);
/* integrate */
PathRadiance L;

7
intern/cycles/kernel/kernel_path_common.h
View File

@@ -19,7 +19,6 @@
CCL_NAMESPACE_BEGIN
ccl_device_inline void kernel_path_trace_setup(KernelGlobals *kg,
ccl_global uint *rng_state,
int sample,
int x, int y,
uint *rng_hash,
@@ -30,11 +29,7 @@ ccl_device_inline void kernel_path_trace_setup(KernelGlobals *kg,
int num_samples = kernel_data.integrator.aa_samples;
if(sample == kernel_data.integrator.start_sample) {
*rng_state = hash_int_2d(x, y);
}
path_rng_init(kg, rng_state, sample, num_samples, rng_hash, x, y, &filter_u, &filter_v);
path_rng_init(kg, sample, num_samples, rng_hash, x, y, &filter_u, &filter_v);
/* sample camera ray */

4
intern/cycles/kernel/kernel_random.h
View File

@@ -15,6 +15,7 @@
*/
#include "kernel/kernel_jitter.h"
#include "util/util_hash.h"
CCL_NAMESPACE_BEGIN
@@ -115,14 +116,13 @@ ccl_device_forceinline void path_rng_2D(KernelGlobals *kg,
}
ccl_device_inline void path_rng_init(KernelGlobals *kg,
ccl_global uint *rng_state,
int sample, int num_samples,
uint *rng_hash,
int x, int y,
float *fx, float *fy)
{
/* load state */
*rng_hash = *rng_state;
*rng_hash = hash_int_2d(x, y);
*rng_hash ^= kernel_data.integrator.seed;
#ifdef __DEBUG_CORRELATION__

1
intern/cycles/kernel/kernel_types.h
View File

@@ -1460,7 +1460,6 @@ typedef struct WorkTile {
uint stride;
ccl_global float *buffer;
ccl_global uint *rng_state;
} WorkTile;
CCL_NAMESPACE_END

2
intern/cycles/kernel/kernels/cpu/kernel_cpu.h
View File

@@ -18,7 +18,6 @@
void KERNEL_FUNCTION_FULL_NAME(path_trace)(KernelGlobals *kg,
float *buffer,
unsigned int *rng_state,
int sample,
int x, int y,
int offset,
@@ -57,7 +56,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
ccl_global void *split_data_buffer,
int num_elements,
ccl_global char *ray_state,
ccl_global uint *rng_state,
int start_sample,
int end_sample,
int sx, int sy, int sw, int sh, int offset, int stride,

4
intern/cycles/kernel/kernels/cpu/kernel_cpu_impl.h
View File

@@ -75,7 +75,6 @@ CCL_NAMESPACE_BEGIN
void KERNEL_FUNCTION_FULL_NAME(path_trace)(KernelGlobals *kg,
float *buffer,
unsigned int *rng_state,
int sample,
int x, int y,
int offset,
@@ -88,7 +87,6 @@ void KERNEL_FUNCTION_FULL_NAME(path_trace)(KernelGlobals *kg,
if(kernel_data.integrator.branched) {
kernel_branched_path_trace(kg,
buffer,
rng_state,
sample,
x, y,
offset,
@@ -97,7 +95,7 @@ void KERNEL_FUNCTION_FULL_NAME(path_trace)(KernelGlobals *kg,
else
# endif
{
kernel_path_trace(kg, buffer, rng_state, sample, x, y, offset, stride);
kernel_path_trace(kg, buffer, sample, x, y, offset, stride);
}
#endif /* KERNEL_STUB */
}

4
intern/cycles/kernel/kernels/cuda/kernel.cu
View File

@@ -42,7 +42,7 @@ kernel_cuda_path_trace(WorkTile *tile, uint total_work_size)
get_work_pixel(tile, work_index, &x, &y, &sample);
KernelGlobals kg;
kernel_path_trace(&kg, tile->buffer, tile->rng_state, sample, x, y, tile->offset, tile->stride);
kernel_path_trace(&kg, tile->buffer, sample, x, y, tile->offset, tile->stride);
}
}
@@ -58,7 +58,7 @@ kernel_cuda_branched_path_trace(WorkTile *tile, uint total_work_size)
get_work_pixel(tile, work_index, &x, &y, &sample);
KernelGlobals kg;
kernel_branched_path_trace(&kg, tile->buffer, tile->rng_state, sample, x, y, tile->offset, tile->stride);
kernel_branched_path_trace(&kg, tile->buffer, sample, x, y, tile->offset, tile->stride);
}
}
#endif

2
intern/cycles/kernel/kernels/cuda/kernel_split.cu
View File

@@ -60,7 +60,6 @@ kernel_cuda_path_trace_data_init(
ccl_global void *split_data_buffer,
int num_elements,
ccl_global char *ray_state,
ccl_global uint *rng_state,
int start_sample,
int end_sample,
int sx, int sy, int sw, int sh, int offset, int stride,
@@ -76,7 +75,6 @@ kernel_cuda_path_trace_data_init(
split_data_buffer,
num_elements,
ray_state,
rng_state,
start_sample,
end_sample,
sx, sy, sw, sh, offset, stride,

3
intern/cycles/kernel/kernels/opencl/kernel.cl
View File

@@ -50,7 +50,6 @@
__kernel void kernel_ocl_path_trace(
ccl_constant KernelData *data,
ccl_global float *buffer,
ccl_global uint *rng_state,
KERNEL_BUFFER_PARAMS,
@@ -68,7 +67,7 @@ __kernel void kernel_ocl_path_trace(
int y = sy + ccl_global_id(1);
if(x < sx + sw && y < sy + sh)
kernel_path_trace(kg, buffer, rng_state, sample, x, y, offset, stride);
kernel_path_trace(kg, buffer, sample, x, y, offset, stride);
}
#else /* __COMPILE_ONLY_MEGAKERNEL__ */

2
intern/cycles/kernel/kernels/opencl/kernel_data_init.cl
View File

@@ -24,7 +24,6 @@ __kernel void kernel_ocl_path_trace_data_init(
ccl_global void *split_data_buffer,
int num_elements,
ccl_global char *ray_state,
ccl_global uint *rng_state,
KERNEL_BUFFER_PARAMS,
int start_sample,
int end_sample,
@@ -41,7 +40,6 @@ __kernel void kernel_ocl_path_trace_data_init(
split_data_buffer,
num_elements,
ray_state,
rng_state,
KERNEL_BUFFER_ARGS,
start_sample,
end_sample,

2
intern/cycles/kernel/kernels/opencl/kernel_split_function.h
View File

@@ -23,7 +23,6 @@ __kernel void KERNEL_NAME_EVAL(kernel_ocl_path_trace, KERNEL_NAME)(
ccl_global void *split_data_buffer,
ccl_global char *ray_state,
ccl_global uint *rng_state,
KERNEL_BUFFER_PARAMS,
@@ -42,7 +41,6 @@ __kernel void KERNEL_NAME_EVAL(kernel_ocl_path_trace, KERNEL_NAME)(
if(ccl_local_id(0) + ccl_local_id(1) == 0) {
kg->data = data;
kernel_split_params.tile.rng_state = rng_state;
kernel_split_params.queue_index = queue_index;
kernel_split_params.use_queues_flag = use_queues_flag;
kernel_split_params.work_pools = work_pools;

6
intern/cycles/kernel/split/kernel_buffer_update.h
View File

@@ -108,10 +108,6 @@ ccl_device void kernel_buffer_update(KernelGlobals *kg,
uint x, y, sample;
get_work_pixel(tile, work_index, &x, &y, &sample);
/* Remap rng_state to current pixel. */
ccl_global uint *rng_state = kernel_split_params.tile.rng_state;
rng_state += tile->offset + x + y*tile->stride;
/* Store buffer offset for writing to passes. */
uint buffer_offset = (tile->offset + x + y*tile->stride) * kernel_data.film.pass_stride;
ccl_global float *buffer = tile->buffer + buffer_offset;
@@ -119,7 +115,7 @@ ccl_device void kernel_buffer_update(KernelGlobals *kg,
/* Initialize random numbers and ray. */
uint rng_hash;
kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng_hash, ray);
kernel_path_trace_setup(kg, sample, x, y, &rng_hash, ray);
if(ray->t != 0.0f) {
/* Initialize throughput, path radiance, Ray, PathState;

15
intern/cycles/kernel/split/kernel_data_init.h
View File

@@ -49,7 +49,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
ccl_global void *split_data_buffer,
int num_elements,
ccl_global char *ray_state,
ccl_global uint *rng_state,
#ifdef __KERNEL_OPENCL__
KERNEL_BUFFER_PARAMS,
@@ -84,7 +83,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
kernel_split_params.tile.offset = offset;
kernel_split_params.tile.stride = stride;
kernel_split_params.tile.rng_state = rng_state;
kernel_split_params.tile.buffer = buffer;
kernel_split_params.total_work_size = sw * sh * num_samples;
@@ -122,7 +120,7 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
*use_queues_flag = 0;
}
/* zero the tiles pixels and initialize rng_state if this is the first sample */
/* zero the tiles pixels if this is the first sample */
if(start_sample == 0) {
int pass_stride = kernel_data.film.pass_stride;
@@ -130,9 +128,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
for(int y = sy; y < sy + sh; y++) {
int index = offset + y * stride;
memset(buffer + (sx + index) * pass_stride, 0, sizeof(float) * pass_stride * sw);
for(int x = sx; x < sx + sw; x++) {
rng_state[index + x] = hash_int_2d(x, y);
}
}
#else
parallel_for(kg, i, sw * sh * pass_stride) {
@@ -146,14 +141,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
*(buffer + index) = 0.0f;
}
parallel_for(kg, i, sw * sh) {
int x = sx + i % sw;
int y = sy + i / sw;
int index = (offset + x + y*stride);
*(rng_state + index) = hash_int_2d(x, y);
}
#endif
}

5
intern/cycles/kernel/split/kernel_path_init.h
View File

@@ -45,10 +45,6 @@ ccl_device void kernel_path_init(KernelGlobals *kg) {
uint x, y, sample;
get_work_pixel(tile, work_index, &x, &y, &sample);
/* Remap rng_state and buffer to current pixel. */
ccl_global uint *rng_state = kernel_split_params.tile.rng_state;
rng_state += tile->offset + x + y*tile->stride;
/* Store buffer offset for writing to passes. */
uint buffer_offset = (tile->offset + x + y*tile->stride) * kernel_data.film.pass_stride;
ccl_global float *buffer = tile->buffer + buffer_offset;
@@ -57,7 +53,6 @@ ccl_device void kernel_path_init(KernelGlobals *kg) {
/* Initialize random numbers and ray. */
uint rng_hash;
kernel_path_trace_setup(kg,
rng_state,
sample,
x, y,
&rng_hash,

11
intern/cycles/render/buffers.cpp
View File

@@ -108,7 +108,6 @@ RenderTile::RenderTile()
stride = 0;
buffer = 0;
rng_state = 0;
buffers = NULL;
}
@@ -131,11 +130,6 @@ void RenderBuffers::device_free()
device->mem_free(buffer);
buffer.clear();
}
if(rng_state.device_pointer) {
device->mem_free(rng_state);
rng_state.clear();
}
}
void RenderBuffers::reset(Device *device, BufferParams& params_)
@@ -149,11 +143,6 @@ void RenderBuffers::reset(Device *device, BufferParams& params_)
buffer.resize(params.width*params.height*params.get_passes_size());
device->mem_alloc("render_buffer", buffer, MEM_READ_WRITE);
device->mem_zero(buffer);
/* allocate rng state */
rng_state.resize(params.width, params.height);
device->mem_alloc("rng_state", rng_state, MEM_READ_WRITE);
}
bool RenderBuffers::copy_from_device(Device *from_device)

3
intern/cycles/render/buffers.h
View File

@@ -74,8 +74,6 @@ public:
/* float buffer */
device_vector<float> buffer;
/* random number generator state */
device_vector<uint> rng_state;
Device *device;
@@ -149,7 +147,6 @@ public:
int tile_index;
device_ptr buffer;
device_ptr rng_state;
RenderBuffers *buffers;

2
intern/cycles/render/session.cpp
View File

@@ -384,7 +384,6 @@ bool Session::acquire_tile(Device *tile_device, RenderTile& rtile)
tile_manager.state.buffer.get_offset_stride(rtile.offset, rtile.stride);
rtile.buffer = buffers->buffer.device_pointer;
rtile.rng_state = buffers->rng_state.device_pointer;
rtile.buffers = buffers;
tile->buffers = buffers;
@@ -442,7 +441,6 @@ bool Session::acquire_tile(Device *tile_device, RenderTile& rtile)
tile->buffers->params.get_offset_stride(rtile.offset, rtile.stride);
rtile.buffer = tile->buffers->buffer.device_pointer;
rtile.rng_state = tile->buffers->rng_state.device_pointer;
rtile.buffers = tile->buffers;
rtile.sample = 0;