From eb293f59f2eb9847b8fd593ac2dde2781ac8ace1 Mon Sep 17 00:00:00 2001 From: Mai Lavelle Date: Thu, 11 May 2017 19:23:49 -0400 Subject: Cycles: Pass all buffers to each kernel call for OpenCL Technically not passing all buffers used by a kernel is undefined behavior. We haven't had any issues with this so far on AMD or Nvidia, but it's known to be a problem with Intel and we received a report from AMD that this is a problem on newer hardware, so we need to make this change at some point. Unfortunately there a cost to being correct, about 5% for the benchmark scenes. For low sample counts it's even worse, I've seen up to 50% slowdown. For the latter case I think adjusting tile updating logic can help, but not sure what that would look like yet (it would be just a few lines change however). --- intern/cycles/kernel/CMakeLists.txt | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'intern/cycles/kernel/CMakeLists.txt') diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt index b85067d4e66..23e9bd311c4 100644 --- a/intern/cycles/kernel/CMakeLists.txt +++ b/intern/cycles/kernel/CMakeLists.txt @@ -122,6 +122,10 @@ set(SRC_KERNELS_CUDA_HEADERS kernels/cuda/kernel_config.h ) +set(SRC_KERNELS_OPENCL_HEADERS + kernels/opencl/kernel_split_function.h +) + set(SRC_CLOSURE_HEADERS closure/alloc.h closure/bsdf.h @@ -452,6 +456,7 @@ add_library(cycles_kernel ${SRC_HEADERS} ${SRC_KERNELS_CPU_HEADERS} ${SRC_KERNELS_CUDA_HEADERS} + ${SRC_KERNELS_OPENCL_HEADERS} ${SRC_BVH_HEADERS} ${SRC_CLOSURE_HEADERS} ${SRC_FILTER_HEADERS} @@ -496,6 +501,7 @@ delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_enqueue_inact delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_next_iteration_setup.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_indirect_subsurface.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_buffer_update.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_split_function.h" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/filter.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/cuda/kernel.cu" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/cuda) delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/cuda/kernel_split.cu" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/cuda) -- cgit v1.2.3