archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity

Fix T92671: confusing Cycles debug logs about CPU architecture

Browse Source 
Instead of printing debug flags listing various CPU and GPU settings that
may or may not be used, print when we are using them. This include CPU
kernel types, OptiX debugging and CUDA and HIP adaptive compilation. BVH
type was already printed.
This commit is contained in:
Brecht Van Lommel
2021-10-31 13:18:28 +01:00
parent 154a060777
commit 806521f703
8 changed files with 19 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
intern/cycles/device/cuda/device_impl.cpp
View File

@@ -378,7 +378,9 @@ string CUDADevice::compile_kernel(const uint kernel_features,
cubin.c_str(),
common_cflags.c_str());
printf("Compiling CUDA kernel ...\n%s\n", command.c_str());
printf("Compiling %sCUDA kernel ...\n%s\n",
(use_adaptive_compilation()) ? "adaptive " : "",
command.c_str());
# ifdef _WIN32
command = "call " + command;
@@ -405,13 +407,15 @@ string CUDADevice::compile_kernel(const uint kernel_features,
bool CUDADevice::load_kernels(const uint kernel_features)
{
/* TODO(sergey): Support kernels re-load for CUDA devices.
/* TODO(sergey): Support kernels re-load for CUDA devices adaptive compile.
*
* Currently re-loading kernel will invalidate memory pointers,
* causing problems in cuCtxSynchronize.
*/
if (cuModule) {
VLOG(1) << "Skipping kernel reload, not currently supported.";
if (use_adaptive_compilation()) {
VLOG(1) << "Skipping CUDA kernel reload for adaptive compilation, not currently supported.";
}
return true;
}