<div dir="ltr"><div>That's the way Intel defined the CPUID bits.</div><div><br></div>I definitely see things like this in the backend<div><br></div><div>let Predicates = [HasBWI, HasVLX]<br></div><div><br></div><div>and the comma there does mean AND so at least some of it is enforced.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Oct 16, 2015 at 12:35 AM, Eric Christopher <span dir="ltr"><<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Also, is the backend enforcing this? I.e. I should be able to construct a testcase that has avx512vl or avx512bw that will fail to select yes?<span class="HOEnZb"><font color="#888888"><div><br></div><div>-eric</div></font></span></div><div class="HOEnZb"><div class="h5"><br><div class="gmail_quote"><div dir="ltr">On Fri, Oct 16, 2015 at 12:34 AM Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Lovely. Cc'ing Justin here. We'll need to figure out some change for these builtins here if we need to have and as well as or.<div><br></div><div>(Why on earth are these builtins subject to two separate features? *sigh* It's ridiculous.)</div><div><br></div><div></div></div><div dir="ltr"><div>-eric</div></div><div dir="ltr"><div><br><br><div class="gmail_quote"><div dir="ltr">On Thu, Oct 15, 2015 at 11:59 PM Craig Topper <<a href="mailto:craig.topper@gmail.com" target="_blank">craig.topper@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Correct you avx512vl means it support 128 and 256-bit vectors. avx512bw means it supports byte and word elements. So you actually need both.</div><div class="gmail_extra"></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Oct 15, 2015 at 11:57 PM, Eric Christopher <span dir="ltr"><<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">The comma separated list is currently represented as "one of these". I couldn't parse your first sentence, for the avx512 ones are you saying that it requires both and not just one of the options?<span><font color="#888888"><div><br></div><div>-eric</div></font></span></div><div><div><br><div class="gmail_quote"><div dir="ltr">On Thu, Oct 15, 2015 at 11:55 PM Craig Topper <<a href="mailto:craig.topper@gmail.com" target="_blank">craig.topper@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">So for the AVX512 ones that list 2 features those features are both required, but for FMA you need either one of the features but not both. What is the comma separated list currently implemented as?</div><div class="gmail_extra"></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Oct 15, 2015 at 3:46 PM, Eric Christopher via cfe-commits <span dir="ltr"><<a href="mailto:cfe-commits@lists.llvm.org" target="_blank">cfe-commits@lists.llvm.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Author: echristo<br>
Date: Thu Oct 15 17:46:02 2015<br>
New Revision: 250470<br>
<br>
URL: <a href="http://llvm.org/viewvc/llvm-project?rev=250470&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project?rev=250470&view=rev</a><br>
Log:<br>
Fix the subtarget features required by some x86 builtins.<br>
<br>
Update the fma builtins to be fma/fma4 until some we can find some<br>
documentation either way.<br>
<br>
Update a couple of the avx intrinsics because they were in the wrong<br>
category.<br>
<br>
Modified:<br>
cfe/trunk/include/clang/Basic/BuiltinsX86.def<br>
<br>
Modified: cfe/trunk/include/clang/Basic/BuiltinsX86.def<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/BuiltinsX86.def?rev=250470&r1=250469&r2=250470&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/BuiltinsX86.def?rev=250470&r1=250469&r2=250470&view=diff</a><br>
==============================================================================<br>
--- cfe/trunk/include/clang/Basic/BuiltinsX86.def (original)<br>
+++ cfe/trunk/include/clang/Basic/BuiltinsX86.def Thu Oct 15 17:46:02 2015<br>
@@ -718,38 +718,38 @@ TARGET_BUILTIN(__builtin_ia32_sha256msg1<br>
TARGET_BUILTIN(__builtin_ia32_sha256msg2, "V4iV4iV4i", "", "sha")<br>
<br>
// FMA<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddss, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddsd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubss, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubsd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddss, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddsd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubss, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubsd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps, "V4fV4fV4fV4f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd, "V2dV2dV2dV2d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddpd256, "V4dV4dV4dV4d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubpd256, "V4dV4dV4dV4d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmaddpd256, "V4dV4dV4dV4d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfnmsubpd256, "V4dV4dV4dV4d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd256, "V4dV4dV4dV4d", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddps256, "V8fV8fV8fV8f", "", "fma")<br>
-TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd256, "V4dV4dV4dV4d", "", "fma")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddss, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddsd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubss, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubsd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddss, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddsd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubss, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubsd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddps, "V4fV4fV4fV4f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd, "V2dV2dV2dV2d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfnmsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmaddsubpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddps256, "V8fV8fV8fV8f", "", "fma,fma4")<br>
+TARGET_BUILTIN(__builtin_ia32_vfmsubaddpd256, "V4dV4dV4dV4d", "", "fma,fma4")<br>
<br>
TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl")<br>
TARGET_BUILTIN(__builtin_ia32_vfmaddpd128_mask3, "V2dV2dV2dV2dUc", "", "avx512vl")<br>
@@ -952,8 +952,8 @@ TARGET_BUILTIN(__builtin_ia32_cmpps256_m<br>
TARGET_BUILTIN(__builtin_ia32_cmpps128_mask, "UcV4fV4fIiUc", "", "avx512vl")<br>
<br>
TARGET_BUILTIN(__builtin_ia32_pcmpeqb512_mask, "LLiV64cV64cLLi", "", "avx512bw")<br>
-TARGET_BUILTIN(__builtin_ia32_pcmpeqd512_mask, "sV16iV16is", "", "avx512bw")<br>
-TARGET_BUILTIN(__builtin_ia32_pcmpeqq512_mask, "cV8LLiV8LLic", "", "avx512bw")<br>
+TARGET_BUILTIN(__builtin_ia32_pcmpeqd512_mask, "sV16iV16is", "", "avx512f")<br>
+TARGET_BUILTIN(__builtin_ia32_pcmpeqq512_mask, "cV8LLiV8LLic", "", "avx512f")<br>
TARGET_BUILTIN(__builtin_ia32_pcmpeqw512_mask, "iV32sV32si", "", "avx512bw")<br>
<br>
TARGET_BUILTIN(__builtin_ia32_pcmpeqb256_mask, "iV32cV32ci", "", "avx512vl,avx512bw")<br>
@@ -966,8 +966,8 @@ TARGET_BUILTIN(__builtin_ia32_pcmpeqq128<br>
TARGET_BUILTIN(__builtin_ia32_pcmpeqw128_mask, "cV8sV8sc", "", "avx512vl,avx512bw")<br>
<br>
TARGET_BUILTIN(__builtin_ia32_pcmpgtb512_mask, "LLiV64cV64cLLi", "", "avx512bw")<br>
-TARGET_BUILTIN(__builtin_ia32_pcmpgtd512_mask, "sV16iV16is", "", "avx512bw")<br>
-TARGET_BUILTIN(__builtin_ia32_pcmpgtq512_mask, "cV8LLiV8LLic", "", "avx512bw")<br>
+TARGET_BUILTIN(__builtin_ia32_pcmpgtd512_mask, "sV16iV16is", "", "avx512f")<br>
+TARGET_BUILTIN(__builtin_ia32_pcmpgtq512_mask, "cV8LLiV8LLic", "", "avx512f")<br>
TARGET_BUILTIN(__builtin_ia32_pcmpgtw512_mask, "iV32sV32si", "", "avx512bw")<br>
<br>
TARGET_BUILTIN(__builtin_ia32_pcmpgtb256_mask, "iV32cV32ci", "", "avx512vl,avx512bw")<br>
<br>
<br>
_______________________________________________<br>
cfe-commits mailing list<br>
<a href="mailto:cfe-commits@lists.llvm.org" target="_blank">cfe-commits@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits</a><br>
</blockquote></div><br><br clear="all"><div><br></div></div><div class="gmail_extra">-- <br><div>~Craig</div>
</div></blockquote></div>
</div></div></blockquote></div><br><br clear="all"><div><br></div></div><div class="gmail_extra">-- <br><div>~Craig</div>
</div></blockquote></div></div></div></blockquote></div>
</div></div></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">~Craig</div>
</div>