<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=us-ascii">
<meta name="Generator" content="Microsoft Word 15 (filtered medium)">
<style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:#0563C1;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:#954F72;
text-decoration:underline;}
p.MsoListParagraph, li.MsoListParagraph, div.MsoListParagraph
{mso-style-priority:34;
margin-top:0in;
margin-right:0in;
margin-bottom:0in;
margin-left:.5in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
span.EmailStyle17
{mso-style-type:personal;
font-family:"Calibri",sans-serif;
color:windowtext;}
span.EmailStyle18
{mso-style-type:personal-reply;
font-family:"Calibri",sans-serif;
color:#1F497D;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@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="#0563C1" vlink="#954F72">
<div class="WordSection1">
<p class="MsoNormal"><a name="_MailEndCompose"><span style="color:#1F497D">Hi Xu,<o:p></o:p></span></a></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D">Thanks for pointing this out and sorry for the delayed response.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal">> It looks when force vectorization is not given, it is not necessary to setup uniform and scalar info for every VF. For a VF, we can do a check before collectUniformsAndScalars() and collectInstsToScalarize() to see if the types used in
the code can actually yield any vector types(after type legalization) or not. If not, there is no point for this VF to participate in VPlan and VF selection. As the (scalar) types can be collected once for all VFs, I guess it is cheap enough. As both collectUniformsAndScalars()
and collectInstsToScalarize() don’t look cheap, doing such check can speed up vectorization, in particular, for large MaxVFs.<o:p></o:p></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D">I’m not sure I understand your proposal. Please, note that uniform values may vary from VF to VF. For example, a branch condition can be uniform (and be kept scalar) for VF=4 but can be divergent for VF=8. For
that reason we have to compute this information for every potential VF since it can be different.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal">Another minor thing is when force vectorization is enabled and MaxVF > 1, expected cost of VF=2 is computed twice at the moment.
<o:p></o:p></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D">Agreed. It would be great if you can submit a patch for that or open a bug. I guess that we just need to introduce a MinVF that is set to 4 when vectorization is forced.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="color:#1F497D">Thanks!<o:p></o:p></span></p>
<div>
<p class="MsoNormal"><span style="color:#1F497D">Diego<o:p></o:p></span></p>
</div>
<p class="MsoNormal"><span style="color:#1F497D"><o:p> </o:p></span></p>
<div>
<div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><a name="_____replyseparator"></a><b>From:</b> llvm-dev [mailto:llvm-dev-bounces@lists.llvm.org]
<b>On Behalf Of </b>Shixiong Xu via llvm-dev<br>
<b>Sent:</b> Friday, June 1, 2018 2:14 PM<br>
<b>To:</b> llvm-dev@lists.llvm.org<br>
<b>Subject:</b> [llvm-dev] [VPlan] about vectorization factor selection<o:p></o:p></p>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Hi,<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Current loop vectorizer uses a range of vectorization factors computed by MaxVF. For each VF, it setups unform and scalar info before building VPlan and the final best VF selection. The best VF is also selected within the VF range.<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal"> for (unsigned VF = 1; VF <= MaxVF; VF *= 2) {<o:p></o:p></p>
<p class="MsoNormal"> // Collect Uniform and Scalar instructions after vectorization with VF.<o:p></o:p></p>
<p class="MsoNormal"> CM.collectUniformsAndScalars(VF);<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal"> // Collect the instructions (and their associated costs) that will be more<o:p></o:p></p>
<p class="MsoNormal"> // profitable to scalarize.<o:p></o:p></p>
<p class="MsoNormal"> if (VF > 1)<o:p></o:p></p>
<p class="MsoNormal"> CM.collectInstsToScalarize(VF);<o:p></o:p></p>
<p class="MsoNormal"> }<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">It looks when force vectorization is not given, it is not necessary to setup uniform and scalar info for every VF. For a VF, we can do a check before collectUniformsAndScalars() and collectInstsToScalarize() to see if the types used in
the code can actually yield any vector types(after type legalization) or not. If not, there is no point for this VF to participate in VPlan and VF selection. As the (scalar) types can be collected once for all VFs, I guess it is cheap enough. As both collectUniformsAndScalars()
and collectInstsToScalarize() don’t look cheap, doing such check can speed up vectorization, in particular, for large MaxVFs.<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Another minor thing is when force vectorization is enabled and MaxVF > 1, expected cost of VF=2 is computed twice at the moment.
<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal"> bool ForceVectorization = Hints->getForce() == LoopVectorizeHints::FK_Enabled;<o:p></o:p></p>
<p class="MsoNormal"> // Ignore scalar width, because the user explicitly wants vectorization.<o:p></o:p></p>
<p class="MsoNormal"> if (ForceVectorization && MaxVF > 1) {<o:p></o:p></p>
<p class="MsoNormal"> Width = 2;<o:p></o:p></p>
<p class="MsoNormal"> Cost = expectedCost(Width).first / (float)Width;<o:p></o:p></p>
<p class="MsoNormal"> }<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal"> for (unsigned i = 2; i <= MaxVF; i *= 2) {<o:p></o:p></p>
<p class="MsoNormal"> // Notice that the vector loop needs to be executed less times, so<o:p></o:p></p>
<p class="MsoNormal"> // we need to divide the cost of the vector loops by the width of<o:p></o:p></p>
<p class="MsoNormal"> // the vector elements.<o:p></o:p></p>
<p class="MsoNormal"> VectorizationCostTy C = expectedCost(i);<o:p></o:p></p>
<p class="MsoNormal"> float VectorCost = C.first / (float)i;<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal"><span lang="EN-IE">Cheers,<o:p></o:p></span></p>
<p class="MsoNormal"><span lang="EN-IE"><o:p> </o:p></span></p>
<p class="MsoNormal"><span lang="EN-IE">Shixiong (Jason) Xu<o:p></o:p></span></p>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
</body>
</html>