[llvm-commits] [PATCH][Review request] add support for AVX enhanced comparison predicates

Kay Tiong Khoo kkhoo at perfwizard.com
Fri Mar 2 13:25:27 PST 2012

Please review the attached patch which adds support for Intel's enhanced FP
comparison predicates that were added with AVX.

The patch includes:
1. Addition of 5-bit immediate field (SSE comparison remains 3-bit) and
AVXCC operand type
2. Fix broken assert statement in "printSSECC()"
3. Fix tablegen definitions for (V)CMP* comparison insts, so AVX variants
have a 5-bit compare immediate
4. A big pile of assembly and disassembly tests

This patch also includes the changes and test to fix disassembly of the
non-AVX CMPPD/CMPPS insts. I submitted a patch for that on Feb 29, but I
don't think anybody has committed it yet. This patch does not distinguish
this feature as Intel-only -- if anyone has advice on how to do that,
please let me know and/or modify the patch. AMD does not support the
enhanced predicates (yet).

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120302/8f241102/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: avx_cmp.diff
Type: application/octet-stream
Size: 47379 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120302/8f241102/attachment.obj>

More information about the llvm-commits mailing list