<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<meta name="Generator" content="Microsoft Word 14 (filtered medium)">
<style><!--
/* Font Definitions */
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
p.MsoAcetate, li.MsoAcetate, div.MsoAcetate
{mso-style-priority:99;
mso-style-link:"Balloon Text Char";
margin:0in;
margin-bottom:.0001pt;
font-size:8.0pt;
font-family:"Tahoma","sans-serif";}
span.BalloonTextChar
{mso-style-name:"Balloon Text Char";
mso-style-priority:99;
mso-style-link:"Balloon Text";
font-family:"Tahoma","sans-serif";}
span.EmailStyle19
{mso-style-type:personal-reply;
font-family:"Courier New";
color:#1F497D;}
.MsoChpDefault
{mso-style-type:export-only;
font-family:"Calibri","sans-serif";}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
</head>
<body lang="EN-US" link="blue" vlink="purple">
<div class="WordSection1">
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">Thanks Sanjay,<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">I'm still digging into PR27681.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">The problem is reproducible without my change, by adding an option like -march=slm to enable the post scheduler.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">It looks to be an issue with the high byteable registers (e.g. %CH) and the antidependency breaker.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">-- mitch<o:p></o:p></span></p>
<p class="MsoNormal"><a name="_MailEndCompose"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"><o:p> </o:p></span></a></p>
<div style="border:none;border-left:solid blue 1.5pt;padding:0in 0in 0in 4.0pt">
<div>
<div style="border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><a name="_____replyseparator"></a><b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">From:</span></b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif""> Sanjay Patel [mailto:spatel@rotateright.com]
<br>
<b>Sent:</b> Wednesday, May 11, 2016 8:44 AM<br>
<b>To:</b> Bodart, Mitch L<br>
<b>Cc:</b> David Majnemer; llvm-commits<br>
<b>Subject:</b> Re: [llvm] r267809 - [X86] Enable the post-RA-scheduler for clang's default 32-bit cpu.<o:p></o:p></span></p>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt">Hi Mitch -<o:p></o:p></p>
</div>
<p class="MsoNormal" style="margin-bottom:12.0pt">I linked a possible duplicate in bugzilla:<br>
<a href="https://llvm.org/bugs/show_bug.cgi?id=27580#c2">https://llvm.org/bugs/show_bug.cgi?id=27580#c2</a><o:p></o:p></p>
</div>
<p class="MsoNormal">The bug manifests when post-RA-scheduler is enabled, but I haven't dug further to see if it's actually a scheduler bug or something else.<o:p></o:p></p>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
</div>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<p class="MsoNormal">On Mon, May 9, 2016 at 11:00 AM, Bodart, Mitch L via llvm-commits <<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a>> wrote:<o:p></o:p></p>
<div>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">Certainly, I'll get right on it.</span><o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"> </span><o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D">-- mitch</span><o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><a name="m_3970184271960693835__MailEndCompose"><span style="font-size:11.0pt;font-family:"Courier New";color:#1F497D"> </span></a><o:p></o:p></p>
<div style="border:none;border-left:solid blue 1.5pt;padding:0in 0in 0in 4.0pt">
<div>
<div style="border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><a name="m_3970184271960693835______replyseparato"></a><b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">From:</span></b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">
David Majnemer [mailto:<a href="mailto:david.majnemer@gmail.com" target="_blank">david.majnemer@gmail.com</a>]
<br>
<b>Sent:</b> Monday, May 09, 2016 9:32 AM<br>
<b>To:</b> Bodart, Mitch L<br>
<b>Cc:</b> llvm-commits<br>
<b>Subject:</b> Re: [llvm] r267809 - [X86] Enable the post-RA-scheduler for clang's default 32-bit cpu.</span><o:p></o:p></p>
</div>
</div>
<div>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">This seems to have caused PR27681 <<a href="http://llvm.org/PR27681" target="_blank">http://llvm.org/PR27681</a>><o:p></o:p></p>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Mitch, can you please take a look?<o:p></o:p></p>
</div>
</div>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">On Wed, Apr 27, 2016 at 3:52 PM, Mitch Bodart via llvm-commits <<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a>> wrote:<o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Author: mbodart<br>
Date: Wed Apr 27 17:52:35 2016<br>
New Revision: 267809<br>
<br>
URL: <a href="http://llvm.org/viewvc/llvm-project?rev=267809&view=rev" target="_blank">
http://llvm.org/viewvc/llvm-project?rev=267809&view=rev</a><br>
Log:<br>
[X86] Enable the post-RA-scheduler for clang's default 32-bit cpu.<br>
<br>
For compilations with no explicit cpu specified, this exhibits<br>
nice gains on Silvermont, with neutral performance on big cores.<br>
<br>
Differential Revision: <a href="http://reviews.llvm.org/D19138" target="_blank">http://reviews.llvm.org/D19138</a><br>
<br>
Added:<br>
llvm/trunk/test/CodeGen/X86/post-ra-sched.ll<br>
Modified:<br>
llvm/trunk/lib/Target/X86/X86.td<br>
llvm/trunk/lib/Target/X86/X86Schedule.td<br>
<br>
Modified: llvm/trunk/lib/Target/X86/X86.td<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86.td?rev=267809&r1=267808&r2=267809&view=diff" target="_blank">
http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86.td?rev=267809&r1=267808&r2=267809&view=diff</a><br>
==============================================================================<br>
--- llvm/trunk/lib/Target/X86/X86.td (original)<br>
+++ llvm/trunk/lib/Target/X86/X86.td Wed Apr 27 17:52:35 2016<br>
@@ -276,12 +276,28 @@ def : Proc<"pentium3", [FeatureX8<br>
FeatureSSE1, FeatureFXSR]>;<br>
def : Proc<"pentium3m", [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
FeatureSSE1, FeatureFXSR, FeatureSlowBTMem]>;<br>
-def : Proc<"pentium-m", [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
- FeatureSSE2, FeatureFXSR, FeatureSlowBTMem]>;<br>
-def : Proc<"pentium4", [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
- FeatureSSE2, FeatureFXSR]>;<br>
-def : Proc<"pentium4m", [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
- FeatureSSE2, FeatureFXSR, FeatureSlowBTMem]>;<br>
+<br>
+// Enable the PostRAScheduler for SSE2 and SSE3 class cpus.<br>
+// The intent is to enable it for pentium4 which is the current default<br>
+// processor in a vanilla 32-bit clang compilation when no specific<br>
+// architecture is specified. This generally gives a nice performance<br>
+// increase on silvermont, with largely neutral behavior on other<br>
+// contemporary large core processors.<br>
+// pentium-m, pentium4m, prescott and nocona are included as a preventative<br>
+// measure to avoid performance surprises, in case clang's default cpu<br>
+// changes slightly.<br>
+<br>
+def : ProcessorModel<"pentium-m", GenericPostRAModel,<br>
+ [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
+ FeatureSSE2, FeatureFXSR, FeatureSlowBTMem]>;<br>
+<br>
+def : ProcessorModel<"pentium4", GenericPostRAModel,<br>
+ [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
+ FeatureSSE2, FeatureFXSR]>;<br>
+<br>
+def : ProcessorModel<"pentium4m", GenericPostRAModel,<br>
+ [FeatureX87, FeatureSlowUAMem16, FeatureMMX,<br>
+ FeatureSSE2, FeatureFXSR, FeatureSlowBTMem]>;<br>
<br>
// Intel Quark.<br>
def : Proc<"lakemont", []>;<br>
@@ -292,10 +308,10 @@ def : ProcessorModel<"yonah", SandyBridg<br>
FeatureFXSR, FeatureSlowBTMem]>;<br>
<br>
// NetBurst.<br>
-def : Proc<"prescott",<br>
- [FeatureX87, FeatureSlowUAMem16, FeatureMMX, FeatureSSE3,<br>
- FeatureFXSR, FeatureSlowBTMem]>;<br>
-def : Proc<"nocona", [<br>
+def : ProcessorModel<"prescott", GenericPostRAModel,<br>
+ [FeatureX87, FeatureSlowUAMem16, FeatureMMX, FeatureSSE3,<br>
+ FeatureFXSR, FeatureSlowBTMem]>;<br>
+def : ProcessorModel<"nocona", GenericPostRAModel, [<br>
FeatureX87,<br>
FeatureSlowUAMem16,<br>
FeatureMMX,<br>
<br>
Modified: llvm/trunk/lib/Target/X86/X86Schedule.td<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Schedule.td?rev=267809&r1=267808&r2=267809&view=diff" target="_blank">
http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Schedule.td?rev=267809&r1=267808&r2=267809&view=diff</a><br>
==============================================================================<br>
--- llvm/trunk/lib/Target/X86/X86Schedule.td (original)<br>
+++ llvm/trunk/lib/Target/X86/X86Schedule.td Wed Apr 27 17:52:35 2016<br>
@@ -633,8 +633,9 @@ def IIC_NOP : InstrItinClass;<br>
// latencies. Since these latencies are not used for pipeline hazards,<br>
// they do not need to be exact.<br>
//<br>
-// The GenericModel contains no instruction itineraries.<br>
-def GenericModel : SchedMachineModel {<br>
+// The GenericX86Model contains no instruction itineraries<br>
+// and disables PostRAScheduler.<br>
+class GenericX86Model : SchedMachineModel {<br>
let IssueWidth = 4;<br>
let MicroOpBufferSize = 32;<br>
let LoadLatency = 4;<br>
@@ -643,6 +644,13 @@ def GenericModel : SchedMachineModel {<br>
let CompleteModel = 0;<br>
}<br>
<br>
+def GenericModel : GenericX86Model;<br>
+<br>
+// Define a model with the PostRAScheduler enabled.<br>
+def GenericPostRAModel : GenericX86Model {<br>
+ let PostRAScheduler = 1;<br>
+}<br>
+<br>
include "X86ScheduleAtom.td"<br>
include "X86SchedSandyBridge.td"<br>
include "X86SchedHaswell.td"<br>
<br>
Added: llvm/trunk/test/CodeGen/X86/post-ra-sched.ll<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/post-ra-sched.ll?rev=267809&view=auto" target="_blank">
http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/post-ra-sched.ll?rev=267809&view=auto</a><br>
==============================================================================<br>
--- llvm/trunk/test/CodeGen/X86/post-ra-sched.ll (added)<br>
+++ llvm/trunk/test/CodeGen/X86/post-ra-sched.ll Wed Apr 27 17:52:35 2016<br>
@@ -0,0 +1,40 @@<br>
+; RUN: llc < %s -mtriple=i386 -mcpu=pentium4 | FileCheck %s<br>
+; RUN: llc < %s -mtriple=i386 -mcpu=pentium4m | FileCheck %s<br>
+; RUN: llc < %s -mtriple=i386 -mcpu=pentium-m | FileCheck %s<br>
+; RUN: llc < %s -mtriple=i386 -mcpu=prescott | FileCheck %s<br>
+; RUN: llc < %s -mtriple=i386 -mcpu=nocona | FileCheck %s<br>
+;<br>
+; Verify that scheduling puts some distance between a load feeding into<br>
+; the address of another load, and that second load. This currently<br>
+; happens during the post-RA-scheduler, which should be enabled by<br>
+; default with the above specified cpus.<br>
+<br>
+@ptrs = external global [0 x i32*], align 4<br>
+@idxa = common global i32 0, align 4<br>
+@idxb = common global i32 0, align 4<br>
+@res = common global i32 0, align 4<br>
+<br>
+define void @addindirect() {<br>
+; CHECK-LABEL: addindirect:<br>
+; CHECK: # BB#0: # %entry<br>
+; CHECK-NEXT: movl idxb, %ecx<br>
+; CHECK-NEXT: movl idxa, %eax<br>
+; CHECK-NEXT: movl ptrs(,%ecx,4), %ecx<br>
+; CHECK-NEXT: movl ptrs(,%eax,4), %eax<br>
+; CHECK-NEXT: movl (%ecx), %ecx<br>
+; CHECK-NEXT: addl (%eax), %ecx<br>
+; CHECK-NEXT: movl %ecx, res<br>
+; CHECK-NEXT: retl<br>
+entry:<br>
+ %0 = load i32, i32* @idxa, align 4<br>
+ %arrayidx = getelementptr inbounds [0 x i32*], [0 x i32*]* @ptrs, i32 0, i32 %0<br>
+ %1 = load i32*, i32** %arrayidx, align 4<br>
+ %2 = load i32, i32* %1, align 4<br>
+ %3 = load i32, i32* @idxb, align 4<br>
+ %arrayidx1 = getelementptr inbounds [0 x i32*], [0 x i32*]* @ptrs, i32 0, i32 %3<br>
+ %4 = load i32*, i32** %arrayidx1, align 4<br>
+ %5 = load i32, i32* %4, align 4<br>
+ %add = add i32 %5, %2<br>
+ store i32 %add, i32* @res, align 4<br>
+ ret void<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" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits</a><o:p></o:p></p>
</div>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
</div>
</div>
</div>
</div>
</div>
</div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><br>
_______________________________________________<br>
llvm-commits mailing list<br>
<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits</a><o:p></o:p></p>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
</div>
</div>
</body>
</html>