<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body dir="auto"><div dir="ltr"></div><div dir="ltr">Great, thank you for getting this started!</div><div dir="ltr"><br>On May 13, 2019, at 7:32 PM, Vitaly Buka <<a href="mailto:vitalybuka@google.com">vitalybuka@google.com</a>> wrote:<br><br></div><blockquote type="cite"><div dir="ltr"><div dir="ltr"><a href="https://reviews.llvm.org/D61879" class="cremed">https://reviews.llvm.org/D61879</a><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr"><strong>From: </strong>Vitaly Buka <span dir="ltr"><<a href="mailto:vitalybuka@google.com">vitalybuka@google.com</a>></span><br><strong>Date: </strong>Mon, May 13, 2019 at 10:23 AM<br><strong>To: </strong>JF Bastien<br><strong>Cc: </strong>Amara Emerson, llvm-dev, Peter Collingbourne<br><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">I have dirty prof-of-concept patch. I am going to rewrite pieces of it during the May starting now.<div><div>Today it's a new pass which does cross-block DSE, module DSE, and global DSE.</div><div>So far the module DSE is the most useful and probably easy integrate to existing DSE.</div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr"><strong>From: </strong>JF Bastien <span dir="ltr"><<a href="mailto:jfbastien@apple.com" target="_blank">jfbastien@apple.com</a>></span><br><strong>Date: </strong>Mon, May 13, 2019 at 9:55 AM<br><strong>To: </strong>Vitaly Buka<br><strong>Cc: </strong>Amara Emerson, llvm-dev, Peter Collingbourne<br><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><br><div><br><blockquote type="cite"><div>On May 10, 2019, at 8:59 PM, Vitaly Buka via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> wrote:</div><br class="gmail-m_-1705651520966340146gmail-m_-4259922444455016778Apple-interchange-newline"><div><div dir="ltr"><div> Sorry for delay, I was busy with other stuff.<br></div><div> CTMark results.</div><div><br></div><div> dse is the current DSE.</div><div> dsem is my experimental module level DSE.</div><div> dsem runs after dse, so it's additionally deleted stores.</div><div><br></div>-O3<br> dse - Number of stores deleted 3033<br> dsem - Number of deleted writes 3148<br><br>-O3 -ftrivial-auto-var-init=pattern<br> dse - Number of stores deleted 5618<br> dsem - Number of deleted writes 3840<div><br></div><div>-O3 -flto<br> dse - Number of stores deleted 3985<br> dsem - Number of deleted writes 3838<br><br>-O3 -flto -ftrivial-auto-var-init=pattern<br> dse - Number of stores deleted 6461<br> dsem - Number of deleted writes 4215<br><br>-Os<br> dse - Number of stores deleted 1443<br> dsem - Number of deleted writes 1517<br><br>-Os -ftrivial-auto-var-init=pattern<br> dse - Number of stores deleted 3951<br> dsem - Number of deleted writes 2259<br><br>-Oz<br> dse - Number of stores deleted 1072<br> dsem - Number of deleted writes 574<br><br>-Oz -ftrivial-auto-var-init=pattern<br> dse - Number of stores deleted 3420<br> dsem - Number of deleted writes 1637<br></div></div></div></blockquote><div><br></div><div>This looks great! Do you have a patch ready to go?</div><div><br></div><br><blockquote type="cite"><div><div class="gmail_quote"><div dir="ltr" class="gmail_attr"><strong>From: </strong>Amara Emerson <span dir="ltr"><<a href="mailto:aemerson@apple.com" target="_blank">aemerson@apple.com</a>></span><br><strong>Date: </strong>Tue, Apr 16, 2019 at 12:10 PM<br><strong>To: </strong>Vitaly Buka<br><strong>Cc: </strong>Alexander Potapenko, llvm-dev, Peter Collingbourne<br><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div>Can you post numbers for how many stores get eliminated from CTMark?<br><div><br><blockquote type="cite"><div>On Apr 16, 2019, at 11:45 AM, Vitaly Buka <<a href="mailto:vitalybuka@google.com" target="_blank">vitalybuka@google.com</a>> wrote:</div><br class="gmail-m_-1705651520966340146gmail-m_-4259922444455016778gmail-m_-6177686787058625214Apple-interchange-newline"><div><div dir="ltr">I tried -Os and effect of new approach significantly increases.<div>I run regular DSE and immediately myDSE. With -Os myDSE removes more than 50% of DSE number.</div><div>Which is expected as -Os inlines less and regular DSE can't remove over function call.</div></div><br><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" target="_blank">glider@google.com</a>> wrote:<br></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>
<<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> wrote:<br>
><br>
><br>
> > On Apr 15, 2019, at 1:51 PM, Vitaly Buka via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> wrote:<br>
> ><br>
> > Hi JF,<br>
> ><br>
> > I've heard that you are interested DSE improvements and maybe we need to be in sync.<br>
> > So far I experimented with following DSE improvements:<br>
> ><br>
> > * Cross-block DSE, it eliminates additional 7% stores comparing to existing DSE. But it's not visible on benchmarks.<br>
> 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>
><br>
> Thanks,<br>
> Amara<br>
> ><br>
> > * Cross-block + Interprocedural analysis to annotate each function argument with:<br>
> > - can read before write<br>
> > - will always write<br>
> > This annotations gets me 20% stores deleted additional to the current DSE.<br>
I believe we can only benefit from removing extra stores.<br>
Hot functions in existing benchmarks are probably optimized good<br>
enough already, but speeding up the long tail is also important.<br>
Also, at least the repro in<br>
<a href="https://bugs.llvm.org/show_bug.cgi?id=40527" rel="noreferrer" target="_blank">https://bugs.llvm.org/show_bug.cgi?id=40527</a> has been extracted from a<br>
real kernel benchmark (hackbench), where this extra store costed us<br>
0.45%<br>
<br>
> > This is on LLVM codebase with -ftrivial-auto-var-init=patter.<br>
> ><br>
> > 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>
> ><br>
> > So now I am also planing to try to extend that to whole program analysis.<br>
> > I will cleanup my code and upload this during this weak, if anyone wants to try.<br>
> ><br>
> > Vitaly.<br>
> > _______________________________________________<br>
> > LLVM Developers mailing list<br>
> > <a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a><br>
> > <a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br>
><br>
> _______________________________________________<br>
> LLVM Developers mailing list<br>
> <a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a><br>
> <a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br>
<br>
<br>
<br>
-- <br>
Alexander Potapenko<br>
Software Engineer<br>
<br>
Google Germany GmbH<br>
Erika-Mann-Straße, 33<br>
80636 München<br>
<br>
Geschäftsführer: Paul Manicle, Halimah DeLaine Prado<br>
Registergericht und -nummer: Hamburg, HRB 86891<br>
Sitz der Gesellschaft: Hamburg<br>
</blockquote></div>
</div></blockquote></div><br></div></blockquote></div>
_______________________________________________<br>LLVM Developers mailing list<br><a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a><br><a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br></div></blockquote></div><br></div></blockquote></div>
</blockquote></div>
</div></blockquote></body></html>