<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">Can you post numbers for how many stores get eliminated from CTMark?<br class=""><div><br class=""><blockquote type="cite" class=""><div class="">On Apr 16, 2019, at 11:45 AM, Vitaly Buka <<a href="mailto:vitalybuka@google.com" class="">vitalybuka@google.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class="">I tried -Os and effect of new approach significantly increases.<div class="">I run regular DSE and immediately myDSE. With -Os myDSE removes more than 50% of DSE number.</div><div class="">Which is expected as -Os inlines less and regular DSE can't remove over function call.</div></div><br class=""><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Apr 16, 2019 at 7:11 AM Alexander Potapenko <<a href="mailto:glider@google.com" class="">glider@google.com</a>> wrote:<br class=""></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On Mon, Apr 15, 2019 at 11:02 PM Amara Emerson via llvm-dev<br class="">
<<a href="mailto:llvm-dev@lists.llvm.org" target="_blank" class="">llvm-dev@lists.llvm.org</a>> wrote:<br class="">
><br class="">
><br class="">
> > On Apr 15, 2019, at 1:51 PM, Vitaly Buka via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank" class="">llvm-dev@lists.llvm.org</a>> wrote:<br class="">
> ><br class="">
> > Hi JF,<br class="">
> ><br class="">
> > I've heard that you are interested DSE improvements and maybe we need to be in sync.<br class="">
> > So far I experimented with following DSE improvements:<br class="">
> ><br class="">
> > * Cross-block DSE, it eliminates additional 7% stores comparing to existing DSE. But it's not visible on benchmarks.<br class="">
> I take it you couldn’t see any runtime impact? If there’s code size improvements that could also be useful, CTMark in the llvm test suite is a useful subset of benchmarks to check this on (as a baseline use -Os to compare code size).<br class="">
><br class="">
> Thanks,<br class="">
> Amara<br class="">
> ><br class="">
> > * Cross-block + Interprocedural analysis to annotate each function argument with:<br class="">
> > - can read before write<br class="">
> > - will always write<br class="">
> > This annotations gets me 20% stores deleted additional to the current DSE.<br class="">
I believe we can only benefit from removing extra stores.<br class="">
Hot functions in existing benchmarks are probably optimized good<br class="">
enough already, but speeding up the long tail is also important.<br class="">
Also, at least the repro in<br class="">
<a href="https://bugs.llvm.org/show_bug.cgi?id=40527" rel="noreferrer" target="_blank" class="">https://bugs.llvm.org/show_bug.cgi?id=40527</a> has been extracted from a<br class="">
real kernel benchmark (hackbench), where this extra store costed us<br class="">
0.45%<br class="">
<br class="">
> > This is on LLVM codebase with -ftrivial-auto-var-init=patter.<br class="">
> ><br class="">
> > As-is it's less than I expected, so I would like to find good benchmark to decide if we should work to make production code from my experiment.<br class="">
> ><br class="">
> > So now I am also planing to try to extend that to whole program analysis.<br class="">
> > I will cleanup my code and upload this during this weak, if anyone wants to try.<br class="">
> ><br class="">
> > Vitaly.<br class="">
> > _______________________________________________<br class="">
> > LLVM Developers mailing list<br class="">
> > <a href="mailto:llvm-dev@lists.llvm.org" target="_blank" class="">llvm-dev@lists.llvm.org</a><br class="">
> > <a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank" class="">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br class="">
><br class="">
> _______________________________________________<br class="">
> LLVM Developers mailing list<br class="">
> <a href="mailto:llvm-dev@lists.llvm.org" target="_blank" class="">llvm-dev@lists.llvm.org</a><br class="">
> <a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank" class="">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br class="">
<br class="">
<br class="">
<br class="">
-- <br class="">
Alexander Potapenko<br class="">
Software Engineer<br class="">
<br class="">
Google Germany GmbH<br class="">
Erika-Mann-Straße, 33<br class="">
80636 München<br class="">
<br class="">
Geschäftsführer: Paul Manicle, Halimah DeLaine Prado<br class="">
Registergericht und -nummer: Hamburg, HRB 86891<br class="">
Sitz der Gesellschaft: Hamburg<br class="">
</blockquote></div>
</div></blockquote></div><br class=""></body></html>