[gmx-developers] OpenACC Development
erik.lindahl at gmail.com
Fri Jul 1 19:46:49 CEST 2016
> On 01 Jul 2016, at 19:42, Millad Ghane <mghane at cs.uh.edu> wrote:
> BTW, PGI is saying
> that they are actually transforming the OpenACC API/Pragmas into CUDA
> code. So, practically, there shouldn't be much more performance loss on
> GPU part.
Based on what?
That’s like saying all C++ implementations of molecular dynamics should have the same performance because it’s the same language.
If that was true, you should not see any performance difference when you disable SIMD. After all, all floating-point math on x86 is implemented with SSE/AVX instructions today.
My prediction is that OpenACC won’t even reach half our native CUDA performance, and then you’ll lose another 30% because PGI completely sucks on the CPU side.
However, feel free to go ahead and prove us wrong ;-)
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the gromacs.org_gmx-developers