<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=Windows-1252">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Hi Eli,</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
> The problem with your proposal, as written, is that the vectorizer is producing the intrinsic. Because we don’t impose any ordering on optimizations before codegen, every optimization pass in LLVM would have to be taught to preserve any @llvm.set.loop.elements.i32
whenever it makes any change. This is completely impractical because the intrinsic isn’t related to anything optimizations would normally look for: it’s a random intrinsic in the middle of nowhere.</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
I do see that point. But is that also not the beauty of it? It just sits in the preheader, if gets removed, then so be it. And if it not recognised, then also no harm done?<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
> Probably the simplest path to get this working is to derive the number of elements in the backend (in HardwareLoops, or your tail predication pass). You should be able to figure it from the masks used in the llvm.masked.load/store instructions in the loop.</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
This is what we are currently doing and works excellent for simpler cases. For the more complicated cases that we now what to handle as well, the pattern matching just becomes a bit too horrible, and it is fragile too. All we need is the information that the
vectoriser already has, and pass this on somehow.</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
As I am really keen to simply our backend pass, would there be another way to pass this information on? If emitting an intrinsic is a blocker, could this be done with a loop annotation?</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Cheers,</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Sjoerd.<br>
</div>
<div id="appendonsend"></div>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Eli Friedman <efriedma@quicinc.com><br>
<b>Sent:</b> 01 May 2020 19:30<br>
<b>To:</b> Sjoerd Meijer <Sjoerd.Meijer@arm.com>; llvm-dev <llvm-dev@lists.llvm.org><br>
<b>Subject:</b> RE: [llvm-dev] LV: predication</font>
<div> </div>
</div>
<style>
<!--
@font-face
{font-family:"Cambria Math"}
@font-face
{font-family:Calibri}
p.x_MsoNormal, li.x_MsoNormal, div.x_MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif}
span.x_EmailStyle19
{font-family:"Calibri",sans-serif;
color:windowtext}
.x_MsoChpDefault
{font-size:10.0pt}
@page WordSection1
{margin:1.0in 1.0in 1.0in 1.0in}
div.x_WordSection1
{}
-->
</style>
<div lang="EN-US" link="#0563C1" vlink="#954F72">
<div class="x_WordSection1">
<p class="x_MsoNormal">The problem with your proposal, as written, is that the vectorizer is producing the intrinsic. Because we don’t impose any ordering on optimizations before codegen, every optimization pass in LLVM would have to be taught to preserve
any @llvm.set.loop.elements.i32 whenever it makes any change. This is completely impractical because the intrinsic isn’t related to anything optimizations would normally look for: it’s a random intrinsic in the middle of nowhere.</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">Probably the simplest path to get this working is to derive the number of elements in the backend (in HardwareLoops, or your tail predication pass). You should be able to figure it from the masks used in the llvm.masked.load/store instructions
in the loop.</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">-Eli</p>
<p class="x_MsoNormal"> </p>
<div>
<div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in">
<p class="x_MsoNormal"><b>From:</b> llvm-dev <llvm-dev-bounces@lists.llvm.org> <b>
On Behalf Of </b>Sjoerd Meijer via llvm-dev<br>
<b>Sent:</b> Friday, May 1, 2020 3:50 AM<br>
<b>To:</b> llvm-dev@lists.llvm.org<br>
<b>Subject:</b> [EXT] [llvm-dev] LV: predication</p>
</div>
</div>
<p class="x_MsoNormal"> </p>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">Hello,</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">We are working on predication for our vector extension (MVE). Since quite a few people are working on predication and different forms of it (e.g. SVE, RISC-V, NEC), I thought I would share what
we would like to add to the loop vectoriser. Hopefully it's just a minor one and not intrusive, but could be interesting and useful for others, and feedback on this is welcome of course.</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">TL;DR:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">We would like the loop vectoriser to emit a new IR intrinsic for certain loops:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> void @llvm.set.loop.elements.i32(i32 )</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">This represents the number of data elements processed by a vector loop, and will be emitted in the preheader block of the vector loop after querying TTI that the backend understands this intrinsic
and that it should be emitted for that loop. The vectoriser patch is available in D79100, and we pick this intrinsic up in the ARM backend here in D79175.</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">Context:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">We are working on predication form that we call tail-predication: a vector hardwareloop has an implicit form of predication that sets active/inactive lanes for the last iteration of the vector
loop. Thus, the scalar epilogue loop (if there is one) is tail-folded and tail-predicated in the main vector body. And to support this, we need to know the number of data elements processed by the loop, which is used in the set up of a tail-predicated vector
loop. This new intrinsic communicates this information from the vectoriser to the codegen passes where we further lower these loops. In our case, we essentially let @llvm.set.loop.elements.i32 emit the trip count of the scalar loop, which represents the number
of data elements processed. Thus, we let the vectoriser emits both the scalar and vector loop trip count.</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">Although in a different stage in the optimisation pipeline, this is exactly what the generic HardwareLoop pass is doing to communicate its information to target specific codegen passes; it emits
a few intrinsics to mark a hardware loop. To illustrate this and also the new intrinsic, this is the flow and life of a tail-predicated vector loop using some heavily edited/reduced examples. First, the vectoriser emits the number of elements processed, and
the loads/stores are masked because tail-folding is applied:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.ph:</span></p>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.set.loop.elements.i32(i32 %N)</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.body:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.masked.store</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br i1 %12, label %.*, label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">After the HardwareLoop pass this is transformed into this, which adds the hardware loop intrinsics:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.ph:</span></p>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.set.loop.elements.i32(i32 %N)</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.set.loop.iterations.i32(i32 %5)</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.body:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.masked.store</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call i32 @llvm.loop.decrement.reg</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br i1 %12, label %.*, label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">We then pick this up in our tail-predication pass, remove @llvm.set.loop.elements intrinsic, and add @vctp which is our intrinsic that generates the mask of active/inactive lanes:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.ph:</span></p>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.set.loop.iterations.i32(i32 %5)</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> vector.body:</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i1> @llvm.arm.mve.vctp32</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call <4 x i32> @llvm.masked.load</span></p>
</div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call void @llvm.masked.store</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> call i32 @llvm.loop.decrement.reg</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> br i1 %12, label %.*, label %vector.body</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">And this is then further lowered to a tail-predicted loop, or reverted to a 'normal' vector loop if some restrictions are not met.</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">Cheers,</span></p>
</div>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">Sjoerd.</span></p>
</div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black"> </span></p>
</div>
</div>
</div>
</body>
</html>