Compiling jacobi-1d with Polly
42 views
Skip to first unread message
Khushboo Chitre
Feb 19, 2022, 1:59:52 AM2/19/22
to Polly Development
Hello,
I’m trying to compile Polybench-c-4.2.1 with the Polly optimizer using the option “-mllvm -polly”. Polly is significantly increasing the execution time of the benchmark "jacobi-1d". The loops in the benchmark are not getting vectorized with Polly, but are getting vectorized without Polly. I tried compiling it with the option “-mllvm -polly -mllvm -polly-vectorizer=polly”. The problem persisted. Am I using the wrong options to compile the benchmark? Suggestions are welcome.
Michael Kruse
Feb 23, 2022, 4:02:00 AM2/23/22
to Khushboo Chitre, Polly Development
jacobi-1d is a benchmark which could profit from time-tiling which the
upstream Polly does not implement. As such, there are not many
optimization opportunities left. However, it still should vectorize.
"-polly-vectorizer=polly" is not such a good vectorizer and I don't
recommend to use it.
You can print the modified loop using "-mllvm -debug-only=polly-ast"
(requires a build of clang with assertions enabled). Also try
"-Rpass-missed=loop-vectorize", it should give you a hint why it did
not vectorize.
Michael
Am Sa., 19. Feb. 2022 um 00:59 Uhr schrieb Khushboo Chitre
<khus...@iiitd.ac.in>:
> You received this message because you are subscribed to the Google Groups "Polly Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to polly-dev+...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/polly-dev/c0f8172b-8eaf-40e7-be72-30b14c8cf2fen%40googlegroups.com.
--
Tardyzentrismus verboten!
upstream Polly does not implement. As such, there are not many
optimization opportunities left. However, it still should vectorize.
"-polly-vectorizer=polly" is not such a good vectorizer and I don't
recommend to use it.
You can print the modified loop using "-mllvm -debug-only=polly-ast"
(requires a build of clang with assertions enabled). Also try
"-Rpass-missed=loop-vectorize", it should give you a hint why it did
not vectorize.
Michael
Am Sa., 19. Feb. 2022 um 00:59 Uhr schrieb Khushboo Chitre
<khus...@iiitd.ac.in>:
>
> Hello,
>
> I’m trying to compile Polybench-c-4.2.1 with the Polly optimizer using the option “-mllvm -polly”. Polly is significantly increasing the execution time of the benchmark "jacobi-1d". The loops in the benchmark are not getting vectorized with Polly, but are getting vectorized without Polly. I tried compiling it with the option “-mllvm -polly -mllvm -polly-vectorizer=polly”. The problem persisted. Am I using the wrong options to compile the benchmark? Suggestions are welcome.
>
>
> --
> Hello,
>
> I’m trying to compile Polybench-c-4.2.1 with the Polly optimizer using the option “-mllvm -polly”. Polly is significantly increasing the execution time of the benchmark "jacobi-1d". The loops in the benchmark are not getting vectorized with Polly, but are getting vectorized without Polly. I tried compiling it with the option “-mllvm -polly -mllvm -polly-vectorizer=polly”. The problem persisted. Am I using the wrong options to compile the benchmark? Suggestions are welcome.
>
>
> You received this message because you are subscribed to the Google Groups "Polly Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to polly-dev+...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/polly-dev/c0f8172b-8eaf-40e7-be72-30b14c8cf2fen%40googlegroups.com.
--
Tardyzentrismus verboten!
Khushboo Chitre
Mar 1, 2022, 10:34:45 PM3/1/22
to Polly Development
Thank you very much!
I will try these suggestions.
Reply all
Reply to author
Forward
0 new messages