<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Generator" content="Microsoft Exchange Server">
<!-- converted from rtf -->
<style><!-- .EmailQuote { margin-left: 1pt; padding-left: 4pt; border-left: #800000 2px solid; } --></style>
</head>
<body>
<font face="Calibri" size="2"><span style="font-size:11pt;">
<div>Hi,</div>
<div> </div>
<div>I think that </div>
<div>def FeatureVectorUAMem : SubtargetFeature<"vector-unaligned-mem",</div>
<div> "HasVectorUAMem", "true",</div>
<div> "Allow unaligned memory operands on vector/SIMD instructions">;</div>
<div> </div>
<div>should be switched-ON on AVX and AVX-512 instructions because:</div>
<div> </div>
<div>According to the AVX spec:</div>
<div>“<font face="Verdana" size="2"><span style="font-size:9pt;">Most arithmetic and data processing instructions encoded using the VEX prefix and</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">performing memory accesses have more flexible memory alignment requirements</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">than instructions that are encoded without the VEX prefix. Specifically,</span></font></div>
<div><font face="Times New Roman" size="4"><span style="font-size:14pt;">• <font face="Verdana" size="2"><span style="font-size:9pt;">With the exception of explicitly aligned 16 or 32 byte SIMD load/store instructions,</span></font></span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">most VEX-encoded, arithmetic and data processing instructions operate in</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">a flexible environment regarding memory address alignment, i.e. VEX-encoded</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">instruction with 32-byte or 16-byte load semantics will support unaligned load</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">operation by default. Memory arguments for most instructions with VEX prefix</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">operate normally without causing #GP(0) on any byte-granularity alignment</span></font></div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">(unlike Legacy SSE instructions).”</span></font></div>
<div> </div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">And the same for AVX-512.</span></font></div>
<div> </div>
<div><font face="Verdana" size="2"><span style="font-size:9pt;">We do not require any alignment while folding loads on the “Peephole Optimizations” on these targets.</span></font></div>
<div> </div>
<div> </div>
<div> </div>
<ul style="margin:0;padding-left:36pt;">
<font face="Times New Roman" size="3" color="#31849B"><span style="font-size:12pt;">
<li><b><i> Elena</i></b></li></span></font>
</ul>
<div> </div>
<div> </div>
<div> </div>
</span></font>
<p>---------------------------------------------------------------------<br>
Intel Israel (74) Limited</p>
<p>This e-mail and any attachments may contain confidential material for<br>
the sole use of the intended recipient(s). Any review or distribution<br>
by others is strictly prohibited. If you are not the intended<br>
recipient, please contact the sender and delete all copies.</p></body>
</html>