R600: Fix min/max matching problems with unordered compares

[Matt Arsenault]

> On Nov 21, 2014, at 5:36 AM, Marek Olšák <maraeo at gmail.com> wrote:
> 
> Hi Matt,
> 
> I've just tested your patches. They don't fix the bug, but there might
> be nothing wrong with them, because I think the legacy opcodes
> themselves are the culprit and switching Mesa to use the mixnum/maxnum
> intrinsics fixes the issue too.
> 
> Marek
> 
> On Thu, Nov 20, 2014 at 12:27 AM, Matt Arsenault <arsenm2 at gmail.com> wrote:
>> These should fix https://bugs.freedesktop.org/show_bug.cgi?id=86432
>> 
>> 
>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 


Here are updated patches which include removing isCommutable from the MachineInstrs

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-R600-SI-Don-t-promote-f32-select-to-i32.patch
Type: application/octet-stream
Size: 4158 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141128/708478db/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-R600-SI-fmin-fmax_legacy-are-not-associative.patch
Type: application/octet-stream
Size: 1135 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141128/708478db/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-R600-Fix-min-max-matching-problems-with-unordered-co.patch
Type: application/octet-stream
Size: 17692 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141128/708478db/attachment-0002.obj>