[llvm-bugs] [Bug 28850] New: AVX512: Invalid operand for {vcvtpd2udqy, vcvtqq2psy, vcvtuqq2psy} instruction

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=28850

            Bug ID: 28850
           Summary: AVX512: Invalid operand for {vcvtpd2udqy, vcvtqq2psy,
                    vcvtuqq2psy} instruction
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: wenzel.jakob at epfl.ch
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

Created attachment 16884
  --> https://llvm.org/bugs/attachment.cgi?id=16884&action=edit
Bzip2 compressed assembly file

Analogous to Bug 28810, I've run into more kinds of situations where number
conversion aliases are missing/not understood by LLVM.

The following lines in particular caused problems:

vcvtpd2udqy     %ymm0, %xmm0
vcvtqq2psy      %ymm0, %xmm0
vcvtuqq2psy     %ymm0, %xmm0

I've added the whole problematic assembly file in case it is useful.

For reference, this is from the test suite of a SIMD vector library which
(amongst other things) exercises many of the different conversions between
vector types. I'm not sure if that's all of them though, there could still be
more missing aliases.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20160804/98827370/attachment.html>