-
Notifications
You must be signed in to change notification settings - Fork 17
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
A100 RuntimeError: CUDA error: no kernel image is available for execution on the device #27
Comments
Sorry, just found this issue for A100. When I followed below instruction and changed setup.py for A100, I hit RuntimeError: Error compiling objects for extension We currently support sm_86 (Ampere, RTX3090/A6000) and sm_89 (Ada, RTX4090). The kernel may run on sm_80 (A100) but expect a significant performance drop. If you want to try it on A100 you could edit setup.py and change arch=compute_86,code=sm_86 to arch=compute_80,code=sm_80. (nunchaku) ubuntu@129-146-61-54:~/nunchaku$ pip install -e . × Building editable for nunchaku (pyproject.toml) did not run successfully.
note: This error originates from a subprocess, and is likely not a problem with pip. |
Looks like NVTX is missing. It should be included in the CUDA Toolkit. Please confirm CUDA is properly installed in your system. Also note that in the current version, the achieved performance on A100 is much lower than the peak performance. One of the main reasons is that the conversion from |
Hi @marvin-0042 I'm also using Lambda Labs NVIDIA A100-SXM4-40GB. However i didn't run the installation in the original env. Instead I run it inside a docker container which comes from this NGC image Hope this can help you! |
thank you so much @dianyo and @sxtyzhangzk !! It works! Lambda Labs NVIDIA A100-SXM4-40GB instructions
|
When run example.py, hit
RuntimeError: CUDA error: no kernel image is available for execution on the device (at /home/ubuntu/nunchaku/src/kernels/awq/gemv_awq.cu:311)
I'm on Lambda A100 GPU instance, unbuntu env. Since env CUDA is 12.4, I changed pip install to use cu124 instead.
pip install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu121
-->
pip install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu124
A100 Unbuntu Env:
PyTorch version: 2.3.1
CUDA available: True
CUDA version: 12.4
Device: NVIDIA A100-SXM4-40GB
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Thu_Mar_28_02:18:24_PDT_2024
Cuda compilation tools, release 12.4, V12.4.131
Build cuda_12.4.r12.4/compiler.34097967_0
python --version
Python 3.10.12
(nunchaku) ubuntu@129-146-61-54:~/nunchaku$ python example.py
RuntimeError: CUDA error: no kernel image is available for execution on the device (at /home/ubuntu/nunchaku/src/kernels/awq/gemv_awq.cu:311)
The text was updated successfully, but these errors were encountered: