[gpaw-users] Issue with CuPy Integration in GPAW for GPU Acceleration

[Mikael Kuisma]

Hi!

It should be now fixed in master. There is also now experimental numpy 2.0 support for CuPy.
https://gitlab.com/gpaw/gpaw/-/merge_requests/2424

BR,
Mikael

________________________________
From: gpaw-users <gpaw-users-bounces at listserv.fysik.dtu.dk> on behalf of Mikael Kuisma via gpaw-users <gpaw-users at listserv.fysik.dtu.dk>
Sent: Sunday, July 28, 2024 12:46 PM
To: gpaw-users at listserv.fysik.dtu.dk <gpaw-users at listserv.fysik.dtu.dk>; David Mauricio Mosquera Narvaez <dmmosqueran at unal.edu.co>
Subject: Re: [gpaw-users] Issue with CuPy Integration in GPAW for GPU Acceleration

Hi!

Thanks for reporting this. It seems that CuPy API has changed from CuPy 12.2 to 13.3.

I created an issue, I can have a look in 1.5 weeks when I get back from vacation, unless somebody has the time to make the change before that (probably we should support both 12.2 and newer, as we still have 12.2 in LUMI).

https://gitlab.com/gpaw/gpaw/-/issues/1226

BR,
Mikael
________________________________
From: gpaw-users <gpaw-users-bounces at listserv.fysik.dtu.dk> on behalf of David Mauricio Mosquera Narvaez via gpaw-users <gpaw-users at listserv.fysik.dtu.dk>
Sent: Sunday, July 28, 2024 5:47 AM
To: gpaw-users at listserv.fysik.dtu.dk <gpaw-users at listserv.fysik.dtu.dk>
Subject: [gpaw-users] Issue with CuPy Integration in GPAW for GPU Acceleration

Hi Gpaw community

I'm a new user of GPAW, and I've been working on setting up GPU acceleration for my calculations. After some effort, I successfully installed GPAW with GPU support in my Conda environment. Below are the details of my setup:


--------------------------------------------------------------------------------------------------------------------------------------------------- | Python Version | 3.9.19 | /home/david-mosquera/miniconda3/envs/gpaw-gpu/bin/python | | GPAW Version | 24.7.0b1 | /home/david-mosquera/gpaw-24.7.0b1/gpaw/ | | ASE Version | 3.23.0 | /home/david-mosquera/miniconda3/envs/gpaw-gpu/lib/python3.9/site-packages/ase/ | | NumPy Version | 1.26.4 | /home/david-mosquera/miniconda3/envs/gpaw-gpu/lib/python3.9/site-packages/numpy/| | SciPy Version | 1.13.1 | /home/david-mosquera/miniconda3/envs/gpaw-gpu/lib/python3.9/site-packages/scipy/| | CuPy Version | 14.0.0a1 | /home/david-mosquera/miniconda3/envs/gpaw-gpu/lib/python3.9/site-packages/cupy/ | | MPI enabled | Yes | | | OpenMP enabled | Yes | | | GPU enabled | Yes | | | GPU-aware MPI | Yes | | | ScaLAPACK enabled | Yes | | | FFTW enabled | Yes | | | PAW-datasets | /home/david-mosquera/gpaw-setups-24.1.0 | | ---------------------------------------------------------------------------------------------------------------------------------------------------

However, when running a GPAW calculation, I encountered the following error:

File "/home/david-mosquera/gpaw-24.7.0b1/gpaw/core/pwacf.py", line 314, in add self.xp.cublas.gemm('N', 'T', File "/home/david-mosquera/miniconda3/envs/gpaw-gpu/lib/python3.9/site-packages/cupy-14.0.0a1-py3.9-linux-x86_64.egg/cupy/__init__.py", line 1087, in __getattr__ raise AttributeError(f"module 'cupy' has no attribute {name!r}") AttributeError: module 'cupy' has no attribute 'cublas'


It seems that CuPy is not recognizing the cublas attribute, despite the CUDA toolkit and libraries being correctly installed and available in the environment. My current environment details are:

  *   NumPy Version: 1.26.4
  *   GPAW Version: 24.7.0b1
  *   ASE Version: 3.23.0
  *   CuPy Version: 14.0.0a1
  *   CUDA Version: 12.5

I'm not sure what might be causing this issue. Any guidance or suggestions to resolve this problem would be greatly appreciated.

Thank you for your assistance.

Best regards,
David Mauricio


Aviso legal: El contenido de este mensaje y los archivos adjuntos son confidenciales y de uso exclusivo de la Universidad Nacional de Colombia. Se encuentran dirigidos sólo para el uso del destinatario al cual van enviados. La reproducción, lectura y/o copia se encuentran prohibidas a cualquier persona diferente a este y puede ser ilegal. Si usted lo ha recibido por error, infórmenos y elimínelo de su correo. Los Datos Personales serán tratados conforme a la Ley 1581 de 2012 y a nuestra Política de Datos Personales que podrá consultar en la página web www.unal.edu.co<http://www.unal.edu.co/>. Las opiniones, informaciones, conclusiones y cualquier otro tipo de dato contenido en este correo electrónico, no relacionados con la actividad de la Universidad Nacional de Colombia, se entenderá como personales y de ninguna manera son avaladas por la Universidad.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20240806/37aaaf40/attachment.htm>