Top Page > Browsing
Date: 2022/08/25 23:31
Name: Mehdi Vejdanihemmat   <>

Dear developers,

Have a nice time;

I noticed that OpenMX has the ability to use GPU instead of CPU, recently. Now my question is, how many times faster will OpenMX run on GPU than CPU?

Thank you in advance,
Mehdi Vejdanihemmat
Page: [1]

Re: GPU vs. CPU ( No.1 )
Date: 2022/08/26 21:55
Name: T. Ozaki


As long as I know, there is only a study which introduced CUDA into OpenMX.
They reported their implementation at
Take a look at Table 5 in the paper. We see that the speedup of Band_DFT_Col by CUDA is 3.78, while that by MPI is 3.73.
The result seemed not to be a driving force towards the support of CUDA in the public version of OpenMX.

Since 2014, GPGPU and related libraries have been continuously being developed, it must be interesting to see if the latest
GPGPU environment really accelerates the OpenMX calculation.


Re: GPU vs. CPU ( No.2 )
Date: 2022/08/26 16:18
Name: Mehdi Vejdanihemmat  <>

Dear Prof. Ozaki,

Thank you very much for your reply. I appreciate it.

Best regards,
Mehdi Vejdanihemmat

Page: [1]

Thread Title (must) Move the thread to the top
Your Name (must)
E-Mail (must)
Password (used in modification of the submitted text)
Comment (must)

   Save Cookie