[LLVMdev] [Polly] GSoC Proposal: Reducing LLVM-Polly Compiling overhead
tanmx_star at yeah.net
Fri May 3 02:39:41 PDT 2013
Thank you very much for your very helpful advice.
Yes, -debug-pass and -time-passes are two very useful and powerful options when evaluating the compile-time of each compiler pass. They are exactly what I need! With these options, I can step into details of the compile-time overhead of each pass. I have finished some preliminary testing based on two randomly selected files from PolyBench and MediaBench. Results are listed on https://gist.github.com/tanstar/5508153 .
Thanks again for your timely advice and help.
At 2013-05-02 17:38:22,"Tobias Grosser" <tobias at grosser.es> wrote:
>On 04/30/2013 04:13 PM, Star Tan wrote:
>> Hi all,
>> How could I find out where the time is spent on between two adjacent Polly passes? Can anyone give me some advice?
>Hi Star Tan,
>I propose to do the performance analysis using the 'opt' tool and
>optimizing LLVM-IR, instead of running it from within clang. For the
>'opt' tool there are two commands that should help you:
>1) -debug-pass=Structure or -debug-pass=Details
>This should give you the list of passes that is executed. You can
>compare the list to see at which point additional passes are scheduled.
>This gives you the time spent in the different passes.
>These two commands may remove the need for a Polly specific profiling
>infrastructure. Also, if you talk about performance issues you see, it
>would be great if you could attach the .ll file you use as well as the
>exact command line you profile.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the llvm-dev