Programming GPUs with Fortran

[@FraserTajima]

Isn’t the call at 16:27 “call saxpykernel …” not call saxpy?

[@farzaddizaji7002]

It was really a nice demonstration for CUDA. Thanks

[@SHARCNET_HPC]

Glad you liked it!

[@captaindunsell8568]

Not necessarily new … we did this with the mainframe and the Cray Vector processors … GPUs are great at vector math and solving simultaneous equations

[@idreeskhan-zp5ey]

Nice explanation. I converted my 2D parallelized fortran code to 3D and now it gives me NaN, any suggestions please?

[@jiangangluo8814]

Does CUDA Fortran work on Nvidia RTX 3080, RTX 3090 graphic card? Or it has to be the GPU processing card, such Tesla, A100? Thank you!

[@kadirgunel5926]

It does not matter, CUDA is developed for all Nvidia cards.

[@snehilsrivastava9204]

what if we are having more than two do loops? since, dimension can be 3, so whether three variables can be assigned with the thread id? and last thing is, suppose we have do k = 1,A_M do i = k+1,A_M beta = 0.0 do j = k, A_N beta = beta + A(j,k)*A(j,i) end do end do end do how come this can be converted into if, as j is depending on k?

[@snehilsrivastava9204]

@@pawelpomorski3711 okay! Thank you.

[@enriqueaguilarolivares9560]

Is there cuda fortran for windows?

[@muhammadfaridkhandaq4379]

what do you mean? The compiler? yes

[@enriqueaguilarolivares9560]

@@muhammadfaridkhandaq4379 thanks