Fix #109550: Cycles: OptiX not able to render without CUDA toolkit
Happens with specific GPUs like NVIDIA A100-SXM4-40GB. They use the compute capability 8.0, which is not explicitly compiled as a cubin, and since7fca0ee76athe PTX is not suitable for it either. The safest solution is to revert the change to a known good state, and re-iterate as needed. Revert "Cycles: Increase the compute model for the PTX kernel" This reverts commit7fca0ee76a. This change would need to be cherry-picked to the 3.6 LTS. Pull Request: https://projects.blender.org/blender/blender/pulls/109636
This commit is contained in:
committed by
Sergey Sharybin
parent
7977247995
commit
03fd794130
@@ -518,7 +518,7 @@ if(NOT APPLE)
|
||||
mark_as_advanced(WITH_CYCLES_DEVICE_CUDA)
|
||||
|
||||
option(WITH_CYCLES_CUDA_BINARIES "Build Cycles NVIDIA CUDA binaries" OFF)
|
||||
set(CYCLES_CUDA_BINARIES_ARCH sm_30 sm_35 sm_37 sm_50 sm_52 sm_60 sm_61 sm_70 sm_75 sm_86 sm_89 compute_89 CACHE STRING "CUDA architectures to build binaries for")
|
||||
set(CYCLES_CUDA_BINARIES_ARCH sm_30 sm_35 sm_37 sm_50 sm_52 sm_60 sm_61 sm_70 sm_75 sm_86 sm_89 compute_75 CACHE STRING "CUDA architectures to build binaries for")
|
||||
option(WITH_CYCLES_CUDA_BUILD_SERIAL "Build cubins one after another (useful on machines with limited RAM)" OFF)
|
||||
option(WITH_CUDA_DYNLOAD "Dynamically load CUDA libraries at runtime (for developers, makes cuda-gdb work)" ON)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user