mirror of
https://github.com/blender/blender
synced 2026-09-29 04:37:17 +03:00
5fc85d1cb7
The new intersect_mnee kernel runs before shade_surface, and shade_surface_mnee is eliminated. That large kernel was causing problems for some GPU compilers. MNEE state is packed into a shadow path state to avoid significantly increasing the path state size. This shadow state is then either turned into an actual shadow ray state or discarded in shade_surface. MNEE was re-enabled on HIP RDNA2 as it works again now. Texture cache misses now also work correctly with MNEE. This adds some extra code to the regular shade_surface kernel even when MNEE is not used, to use the MNEE sampled point instead of sampling a light. But there seems to be no significant performance impact. Co-authored-by: Sergey Sharybin <sergey@blender.org> Pull Request: https://projects.blender.org/blender/blender/pulls/158698
56 lines
1.4 KiB
C++
56 lines
1.4 KiB
C++
/* SPDX-FileCopyrightText: 2011-2022 Blender Foundation
|
|
*
|
|
* SPDX-License-Identifier: Apache-2.0 */
|
|
|
|
#ifdef WITH_CUDA
|
|
|
|
# include "device/cuda/kernel.h"
|
|
# include "device/cuda/device_impl.h"
|
|
|
|
CCL_NAMESPACE_BEGIN
|
|
|
|
void CUDADeviceKernels::load(CUDADevice *device)
|
|
{
|
|
CUmodule cuModule = device->cuModule;
|
|
|
|
for (int i = 0; i < (int)DEVICE_KERNEL_NUM; i++) {
|
|
CUDADeviceKernel &kernel = kernels_[i];
|
|
|
|
if (!device_kernel_has_gpu_function((DeviceKernel)i)) {
|
|
continue;
|
|
}
|
|
|
|
const std::string function_name = std::string("kernel_gpu_") +
|
|
device_kernel_as_string((DeviceKernel)i);
|
|
cuda_device_assert(device,
|
|
cuModuleGetFunction(&kernel.function, cuModule, function_name.c_str()));
|
|
|
|
if (kernel.function) {
|
|
cuda_device_assert(device, cuFuncSetCacheConfig(kernel.function, CU_FUNC_CACHE_PREFER_L1));
|
|
|
|
cuda_device_assert(
|
|
device,
|
|
cuOccupancyMaxPotentialBlockSize(
|
|
&kernel.min_blocks, &kernel.num_threads_per_block, kernel.function, nullptr, 0, 0));
|
|
}
|
|
else {
|
|
LOG_ERROR << "Unable to load kernel " << function_name;
|
|
}
|
|
}
|
|
|
|
loaded = true;
|
|
}
|
|
|
|
const CUDADeviceKernel &CUDADeviceKernels::get(DeviceKernel kernel) const
|
|
{
|
|
return kernels_[(int)kernel];
|
|
}
|
|
|
|
bool CUDADeviceKernels::available(DeviceKernel kernel) const
|
|
{
|
|
return kernels_[(int)kernel].function != nullptr;
|
|
}
|
|
|
|
CCL_NAMESPACE_END
|
|
|
|
#endif /* WITH_CUDA */
|