vLAR-group/GrowSP

train_S3DIS.py error

Closed this issue · 8 comments

Hi! Thanks for sharing your code.
I tried it for S3DIS dataset and am facing the following error when I ran the CUDA_VISIBLE_DEVICES=0, python train_S3DIS.py command:
RuntimeError: CUDA error: CUBLAS_STATUS_INVALID_VALUE when calling cublasSgemm( handle, opa, opb, m, n, k, &alpha, a, lda, b, ldb, &beta, c, ldc).

GrowSPerror

Thank you in advance for your help.

@narges-tk, Hi, seems the error occurred in MinkowskiEngine, did you keep the same virtual environment with us?

Yes, I have created the same virtual environment

Hi, @narges-tk, I guess the MinkowskiEngine may have some errors, you can try another installation:
Please install pytorch and other packages as illustrated in Readme. and activate the environment, then type the follow command to install MinkowskiEngine:

git clone --recursive "https://github.com/NVIDIA/MinkowskiEngine"
cd MinkowskiEngine
git checkout 02fc608bea4c0549b0a7b00ca1bf15dee4a0b228
python setup.py install --force_cuda --blas=openblas 

Many thanks for your help! Now cuda error is resolved.
However, I am getting new following error related to numpy. I did not managed to fix it. Any suggestion is welcome.
Thanks in advance
1
2

I have solved the numpy error. However, I am facing the same cuda error again! Any help is greatly appreciated.
1

coqui-ai/TTS#1517, this issue may provide some help, typing "unset LD_LIBRARY_PATH" in your python environment as coqui-ai/TTS#1517 (comment)

Thanks, it is now resolved!

谢谢,现在已经解决了!

I have the same problem, can you provide a solution please?