<div dir="ltr">Hi Mehdi,<div><br></div><div>I've reverted this in r249528 as it was causing failures in all our internal builders and here too: <span style="line-height:1.5"><a href="http://lab.llvm.org:8011/builders/clang-native-arm-lnt/builds/14453">http://lab.llvm.org:8011/builders/clang-native-arm-lnt/builds/14453</a></span></div><div><span style="line-height:1.5"><br></span></div><div><span style="line-height:1.5">Apologies,</span></div><div><span style="line-height:1.5"><br></span></div><div><span style="line-height:1.5">James</span></div>
</div><br><div class="gmail_quote"><div dir="ltr">On Tue, 6 Oct 2015 at 18:21 Mehdi Amini via llvm-commits <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Author: mehdi_amini<br>
Date: Tue Oct 6 12:19:20 2015<br>
New Revision: 249431<br>
<br>
URL: <a href="http://llvm.org/viewvc/llvm-project?rev=249431&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project?rev=249431&view=rev</a><br>
Log:<br>
This patch builds on top of D13378 to handle constant condition.<br>
<br>
With this patch, clang -O3 optimizes correctly providing > 1000x speedup on this artificial benchmark):<br>
<br>
for (a=0; a<n; a++)<br>
for (b=0; b<n; b++)<br>
for (c=0; c<n; c++)<br>
for (d=0; d<n; d++)<br>
for (e=0; e<n; e++)<br>
for (f=0; f<n; f++)<br>
x++;<br>
>From test-suite/SingleSource/Benchmarks/Shootout/nestedloop.c<br>
<br>
Reviewers: sanjoyd<br>
<br>
Differential Revision: <a href="http://reviews.llvm.org/D13390" rel="noreferrer" target="_blank">http://reviews.llvm.org/D13390</a><br>
<br>
From: Mehdi Amini <<a href="mailto:mehdi.amini@apple.com" target="_blank">mehdi.amini@apple.com</a>><br>
<br>
Added:<br>
llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll<br>
Modified:<br>
llvm/trunk/lib/Analysis/ScalarEvolution.cpp<br>
<br>
Modified: llvm/trunk/lib/Analysis/ScalarEvolution.cpp<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/ScalarEvolution.cpp?rev=249431&r1=249430&r2=249431&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/ScalarEvolution.cpp?rev=249431&r1=249430&r2=249431&view=diff</a><br>
==============================================================================<br>
--- llvm/trunk/lib/Analysis/ScalarEvolution.cpp (original)<br>
+++ llvm/trunk/lib/Analysis/ScalarEvolution.cpp Tue Oct 6 12:19:20 2015<br>
@@ -3904,6 +3904,11 @@ const SCEV *ScalarEvolution::createNodeF<br>
Value *Cond,<br>
Value *TrueVal,<br>
Value *FalseVal) {<br>
+ // Handle "constant" branch or select. This can occur for instance when a<br>
+ // loop pass transforms an inner loop and moves on to process the outer loop.<br>
+ if (auto *CI = dyn_cast<ConstantInt>(Cond))<br>
+ return getSCEV(CI->isOne() ? TrueVal : FalseVal);<br>
+<br>
// Try to match some simple smax or umax patterns.<br>
auto *ICI = dyn_cast<ICmpInst>(Cond);<br>
if (!ICI)<br>
<br>
Added: llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll?rev=249431&view=auto" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll?rev=249431&view=auto</a><br>
==============================================================================<br>
--- llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll (added)<br>
+++ llvm/trunk/test/Analysis/ScalarEvolution/constant_condition.ll Tue Oct 6 12:19:20 2015<br>
@@ -0,0 +1,51 @@<br>
+; RUN: opt -analyze -scalar-evolution < %s | FileCheck %s<br>
+<br>
+define i32 @branch_true(i32 %x, i32 %y) {<br>
+; CHECK-LABEL: Classifying expressions for: @branch_true<br>
+ entry:<br>
+ br i1 true, label %add, label %merge<br>
+<br>
+ add:<br>
+ %sum = add i32 %x, %y<br>
+ br label %merge<br>
+<br>
+ merge:<br>
+ %v = phi i32 [ %sum, %add ], [ %x, %entry ]<br>
+; CHECK: %v = phi i32 [ %sum, %add ], [ %x, %entry ]<br>
+; CHECK-NEXT: --> (%x + %y) U: full-set S: full-set<br>
+ ret i32 %v<br>
+}<br>
+<br>
+define i32 @branch_false(i32 %x, i32 %y) {<br>
+; CHECK-LABEL: Classifying expressions for: @branch_false<br>
+ entry:<br>
+ br i1 false, label %add, label %merge<br>
+<br>
+ add:<br>
+ %sum = add i32 %x, %y<br>
+ br label %merge<br>
+<br>
+ merge:<br>
+ %v = phi i32 [ %sum, %add ], [ %x, %entry ]<br>
+; CHECK: %v = phi i32 [ %sum, %add ], [ %x, %entry ]<br>
+; CHECK-NEXT: --> %x U: full-set S: full-set<br>
+ ret i32 %v<br>
+}<br>
+<br>
+define i32 @select_true(i32 %x, i32 %y) {<br>
+; CHECK-LABEL: Classifying expressions for: @select_true<br>
+ entry:<br>
+ %v = select i1 true, i32 %x, i32 %y<br>
+; CHECK: %v = select i1 true, i32 %x, i32 %y<br>
+; CHECK-NEXT: --> %x U: full-set S: full-set<br>
+ ret i32 %v<br>
+}<br>
+<br>
+define i32 @select_false(i32 %x, i32 %y) {<br>
+; CHECK-LABEL: Classifying expressions for: @select_false<br>
+ entry:<br>
+ %v = select i1 false, i32 %x, i32 %y<br>
+; CHECK: %v = select i1 false, i32 %x, i32 %y<br>
+; CHECK-NEXT: --> %y U: full-set S: full-set<br>
+ ret i32 %v<br>
+}<br>
<br>
<br>
_______________________________________________<br>
llvm-commits mailing list<br>
<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits</a><br>
</blockquote></div>