Wants some advice about using OpenACC to modify application

Accelerated Computing HPC Compilers Legacy PGI Compilers
1

Hi all,
I want to improve HPC application performance, like HPCC.
First, I try to rewrite HPL source code under folder $(HPL_dir)/src/
But when I finished it, I found that I can’t use pgprof to analysis my code to confirm if I get a better performance than before without executable file .
Even though I use make file to build it (I haven’t finished yet because of another problem), I only get one executable file.
I don’t think one executable file can help me to analysis many code (more then one) behind it.
Did anyone ever use PGI compiler to modify application and get better performance? I really want to know how can I analysis my code, or other method to help me confirm new code performance.

Thanks.

2

Hi alan0415,

Why can’t you use pgprof?

Are you using it and just not seeing any speed-up?

Did anyone ever use PGI compiler to modify application and get better performance?

Yes, I do this quite frequently. Though for the question as to how, it’s a bit too big of a topic to cover in a UF post. If you can formulate a more specific question, I’ll do my best to answer.

Have you taken any training on using OpenACC? If not, I’d suggest starting by looking at the material OpenACC’s resources page https://www.openacc.org/resources

For pgprof, please see our documentation: Profiler Guide :: PGI version 19.1 Documentation for x86 and NVIDIA Processors

-Mat