-1

I have a C++ code and I used CBLAS to compute dgemm and dtrsm. I am interested to use GPUs for performance. With the tests I have done I could use NVBLAS using nvblas.h, however it is not close to CBLAS. I think I can change my code to call dgemm just like nvblas.h. Is there an easier way to link CBLAS using NVBLAS?

Aznaveh
  • 558
  • 8
  • 27

1 Answers1

0

https://docs.nvidia.com/cuda/nvblas/index.html#configuration Because NVBLAS is a drop-in replacement of BLAS, it must be configured through an ASCII text file that describes how many and which GPUs can participate in the intercepted BLAS calls. The configuration file is parsed at the time of the loading of the library. The format of the configuration file is based on keywords optionally followed by one or more user-defined parameters. At most one keyword per line is allowed. Blank lines or lines beginning with the character # are ignored.

王保罗
  • 1
  • 2
    Your answer could be improved with additional supporting information. Please [edit] to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers [in the help center](/help/how-to-answer). – Community May 31 '23 at 08:24