DP Cublas functions

Hi there,

I was just looking over this tread: https://forums.developer.nvidia.com/t/about-cublas/133082/1 as well as looking at the cublasTestSgemm.F90 example in the /pgi directory.

Anyway, I was just fiddling around with the example and I noticed that it doesn’t recognise dgemm (i.e., if I just switch sgemm to dgemm in the example). Are the double precision functions not supported? Is there a list some where containing the functions in the cublas module?


Hi Rob,

DGEMM is supported, but I’m wondering if you changed all the arguments to use double precision? You can instead use the flag “-r8” to change the default kind of reals to double precision.

  • Mat