-
Notifications
You must be signed in to change notification settings - Fork 370
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CUDA ARCH 9.0 unsupported? #3157
Comments
In CMakeLists.txt: We remove certain architectures that are not supportedset(UNSUPPORTED_CUDA_ARCHITECTURES_TORCH 90) |
:( |
Hi @CHNtentes - we are working to add this back. Some of the executors failed with arch 90 and was causing build instability. PR should be up soon. |
Thanks for your reply. Do I need to manually clone & build after PR is up, or you will release a 1.33.1 version? |
@CHNtentes We will go ahead and provide a release once we get this completed, thanks! |
Hi, I saw that you merged a PR about this issue. Could you share when you will release a fix version? Our H20 server has not been in use for long time. |
(qnn) root@gpu_h20:/nvme/shanghai-2/ltg/aimet/build# cmake .. -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DENABLE_CUDA=ON -DENABLE_TORCH=ON -DENABLE_TENSORFLOW=ON -DENABLE_ONNX=ON
-- The C compiler identification is GNU 9.4.0
-- The CXX compiler identification is GNU 9.4.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/c++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Set SW_VERSION = 1.32.0 from /nvme/shanghai-2/ltg/aimet/packaging/version.txt
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE
Compiling with CUDA enabled
-- Found CUDAToolkit: /usr/local/cuda/targets/x86_64-linux/include (found version "11.8.89")
-- ** Initial CMAKE_CUDA_ARCHITECTURES = 52;60;61;70;72 **
-- The CUDA compiler identification is NVIDIA 11.8.89
-- Detecting CUDA compiler ABI info
-- Detecting CUDA compiler ABI info - done
-- Check for working CUDA compiler: /usr/local/cuda/bin/nvcc - skipped
-- Detecting CUDA compile features
-- Detecting CUDA compile features - done
-- Found CUDA toolkit version 11.8.89, using cu118
-- Found Python3: /root/miniconda3/envs/qnn/bin/python3.10 (found version "3.10.0") found components: Interpreter Development Development.Module Development.Embed
Found python: TRUE, at /root/miniconda3/envs/qnn/lib/libpython3.10.so
-- Found PkgConfig: /usr/bin/pkg-config (found version "0.29.1")
-- Checking for one of the modules 'lapacke'
-- Checking for one of the modules 'opencv'
-- Found ONNX version: 1.14.1
-- AIMET - ENABLE_TORCH: ON
-- Found Torch version: 2.1.2+cu118
-- ** Updated CMAKE_CUDA_ARCHITECTURES to 50;60;70;75;80;86;37;90 **
-- ** Removed unsupported archs (90), Now CMAKE_CUDA_ARCHITECTURES = 50;60;70;75;80;86;37 **
-- ** Updated TORCH_CUDA_ARCH_LIST to 5.0 6.0 7.0 7.5 8.0 8.6 3.7 **
The text was updated successfully, but these errors were encountered: