<div dir="auto">Oh, didn't realize we had this already! Thanks.<div dir="auto">Then, yes, sure. </div></div><div class="gmail_extra"><br><div class="gmail_quote">On Feb 19, 2017 13:39, "Mehdi AMINI via Phabricator" <<a href="mailto:reviews@reviews.llvm.org">reviews@reviews.llvm.org</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">mehdi_amini added inline comments.<br>
<br>
<br>
================<br>
Comment at: docs/LangRef.rst:11655<br>
+reduction of a vector, returning the result as a scalar. The return type matches<br>
+the element-type of the vector input.<br>
+<br>
----------------<br>
mkuper wrote:<br>
> mehdi_amini wrote:<br>
> > delena wrote:<br>
> > > mehdi_amini wrote:<br>
> > > > I think it'd be important to clarify the ordering: do we require a "fast" FMF to change it? If not, why?<br>
> > > I think that in the case of FP operations we need two intrinsics - "ordered" and "fast".<br>
> > Why don't reusing the existing FMF? The intrinsic without FMF would be equivalent to "ordered" and the "fast" FMF flag would allow any ordering.<br>
> I don't think we can currently attach FMF to CallInsts.<br>
We do actually: <a href="http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151214/319516.html" rel="noreferrer" target="_blank">http://lists.llvm.org/<wbr>pipermail/llvm-commits/Week-<wbr>of-Mon-20151214/319516.html</a><br>
<br>
<br>
Repository:<br>
  rL LLVM<br>
<br>
<a href="https://reviews.llvm.org/D30086" rel="noreferrer" target="_blank">https://reviews.llvm.org/<wbr>D30086</a><br>
<br>
<br>
<br>
</blockquote></div></div>