<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">Hi,<div class=""><br class=""></div><div class="">Yes SCEV is pretty limited on this aspect.</div><div class="">This kind of code can trigger LLVM to explode in time/memory: <a href="https://llvm.org/bugs/show_bug.cgi?id=18606" class="">https://llvm.org/bugs/show_bug.cgi?id=18606</a></div><div class=""><br class=""></div><div class="">See also this llvm-dev thread: SCEV implementation and limitations, do we need "pow"? : <a href="http://lists.llvm.org/pipermail/llvm-dev/2014-February/070062.html" class="">http://lists.llvm.org/pipermail/llvm-dev/2014-February/070062.html</a></div><div class=""><br class=""></div><div class="">CC: Sanjoy who may have an opinion on how to improve SCEV for this?</div><div class=""><br class=""></div><div class="">-- </div><div class="">Mehdi</div><div class=""><br class=""></div><div class=""><br class=""><div><blockquote type="cite" class=""><div class="">On Mar 1, 2016, at 11:31 PM, Hongbin Zheng via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" class="">llvm-dev@lists.llvm.org</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class="">if you replace "<span style="font-size:12.8px" class="">zero *= a;</span>" by "<span style="font-size:12.8px" class="">zero += a;</span>", you get:<div class=""><br class=""></div><div class=""><div class="">; Function Attrs: norecurse nounwind readnone uwtable</div><div class="">define i32 @foo(i32 %a) #0 {</div><div class="">entry:</div><div class=""> %0 = mul i32 %a, 10000</div><div class=""> ret i32 %0</div><div class="">}</div></div><div class=""><br class=""></div><div class="">I think the problem is ScalarEvolution only have "SCEVAddRecExpr" for <span style="font-size:12.8px" class="">zero += a, but no corresponding expression to represent and optimize zero *= a;</span></div><div class=""><span style="font-size:12.8px" class=""><br class=""></span></div><div class=""><span style="font-size:12.8px" class=""><br class=""></span></div></div><div class="gmail_extra"><br class=""><div class="gmail_quote">On Wed, Mar 2, 2016 at 3:24 PM, zet via llvm-dev <span dir="ltr" class=""><<a href="mailto:llvm-dev@lists.llvm.org" target="_blank" class="">llvm-dev@lists.llvm.org</a>></span> wrote:<br class=""><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr" class=""><div class="">test.c :</div><div class="">-----------------------------------------------------------------------</div><div class=""><div class="">int foo(int a)</div><div class="">{</div><div class=""> int zero = 0;</div><div class=""> for (int i = 0; i < 10000; i++)</div><div class=""> zero *= a;</div><div class=""> return zero;</div><div class="">}</div></div><div class="">-------------------------------------------------------------------------</div><div class=""><br class=""></div><div class="">run clang : clang -O2 -S test.c -o test.s</div><div class=""><br class=""></div><div class="">My clang version is 3.7.1.</div><div class="">We will get a horrible assembly output.<br class=""></div><div class=""><br class=""></div><div class="">Why constant propagation and other optimization skills can not find out that variable zero is initialized 0, and the only statement in for loop (i.e. zero *= a) always get a 0?</div><div class=""><br class=""></div><div class="">I can read the clang/llvm source, Please tell me more details.</div><div class=""><br class=""></div><div class="">THX</div><span class="HOEnZb"><font color="#888888" class=""><div class=""><br class=""></div><div class=""><br class=""></div>-- <br class=""><div class=""><div dir="ltr" class=""><div class=""><font face="nsimsun, monospace" class=""><span style="color:rgb(51,51,51);line-height:24px;white-space:pre-wrap;background-color:rgb(241,254,221)" class="">业精于勤</span><span style="color:rgb(51,51,51);line-height:24px;white-space:pre-wrap;background-color:rgb(241,254,221)" class="">,</span><span style="color:rgb(51,51,51);line-height:24px;white-space:pre-wrap;background-color:rgb(241,254,221)" class="">荒于嬉..</span></font></div></div></div>
</font></span></div>
<br class="">_______________________________________________<br class="">
LLVM Developers mailing list<br class="">
<a href="mailto:llvm-dev@lists.llvm.org" class="">llvm-dev@lists.llvm.org</a><br class="">
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank" class="">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br class="">
<br class=""></blockquote></div><br class=""></div>
_______________________________________________<br class="">LLVM Developers mailing list<br class=""><a href="mailto:llvm-dev@lists.llvm.org" class="">llvm-dev@lists.llvm.org</a><br class="">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev<br class=""></div></blockquote></div><br class=""></div></body></html>