blender
/
blender
133
398
Fork 579
Code Issues 6.3k Pull Requests 752 Projects 19 Wiki Activity

Merge branch 'master' into blender2.8

This commit is contained in:
Sergey Sharybin 2017-05-02 15:29:00 +02:00
parent ec9ce9e118 4384a7cf46
commit 7f833c0da8
2 changed files with 13 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CMakeLists.txt
View File

@ -369,7 +369,7 @@ mark_as_advanced(WITH_LIBMV_SCHUR_SPECIALIZATIONS)
# Logging/unbit test libraries.
option(WITH_SYSTEM_GFLAGS "Use system-wide Gflags instead of a bundled one" OFF)
option(WITH_SYSTEM_GFLOG "Use system-wide Glog instead of a bundled one" OFF)
option(WITH_SYSTEM_GLOG "Use system-wide Glog instead of a bundled one" OFF)
mark_as_advanced(WITH_SYSTEM_GFLAGS)
mark_as_advanced(WITH_SYSTEM_GLOG)

16
intern/cycles/device/device_cuda.cpp
View File

@ -119,6 +119,7 @@ public:
int cuDevId;
int cuDevArchitecture;
bool first_error;
CUDASplitKernel *split_kernel;
struct PixelMem {
GLuint cuPBO;
@ -221,6 +222,8 @@ public:
cuDevice = 0;
cuContext = 0;
split_kernel = NULL;
need_bindless_mapping = false;
/* intialize */
@ -260,6 +263,8 @@ public:
{
task_pool.stop();
delete split_kernel;
if(info.has_bindless_textures) {
tex_free(bindless_mapping);
}
@ -1357,12 +1362,14 @@ public:
requested_features.max_closure = 64;
}
CUDASplitKernel split_kernel(this);
split_kernel.load_kernels(requested_features);
if(split_kernel == NULL) {
split_kernel = new CUDASplitKernel(this);
split_kernel->load_kernels(requested_features);
}
while(task->acquire_tile(this, tile)) {
device_memory void_buffer;
split_kernel.path_trace(task, tile, void_buffer, void_buffer);
split_kernel->path_trace(task, tile, void_buffer, void_buffer);
task->release_tile(tile);
@ -1648,7 +1655,8 @@ int2 CUDASplitKernel::split_kernel_global_size(device_memory& kg, device_memory&
<< string_human_readable_size(free) << ").";
size_t num_elements = max_elements_for_max_buffer_size(kg, data, free / 2);
int2 global_size = make_int2(round_down((int)sqrt(num_elements), 32), (int)sqrt(num_elements));
size_t side = round_down((int)sqrt(num_elements), 32);
int2 global_size = make_int2(side, round_down(num_elements / side, 16));
VLOG(1) << "Global size: " << global_size << ".";
return global_size;
}