[Bug 92595] [HSW,BDW,SKL][GLES 3.1 CTS] Big difference in the results for the ES31-CTS.shader_bitfield_operation.* tests

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Fri Dec 18 07:21:40 PST 2015


https://bugs.freedesktop.org/show_bug.cgi?id=92595

Ian Romanick <idr at freedesktop.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |currojerez at riseup.net,
                   |                            |idr at freedesktop.org,
                   |                            |mattst88 at gmail.com

--- Comment #7 from Ian Romanick <idr at freedesktop.org> ---
(In reply to Marta Löfstedt from comment #6)
> Created attachment 120561 [details]
> Hack that fix the difference in results between HSW and BDW/SKL
> 
> Sorry, I attached the wrong version of the patch and did the same mistake to
> myself when testing on SKL. It is when I hack BRW_REGISTER_TYPE_UW for both
> signed and unsigned that I pass the imulExtended shader_bitfield tests on
> BDW and SKL(kernel 4.2.0).

This code was originally added by 3b48a0ee (authored by Curro, reviewed by
Matt).

For Gen8, it's not obvious to me why there is any modification of the source
types of the MUL.  It doesn't help that the BDW public PRM has *no*
documentation for the MUL or MACH instructions.

I'm also a little disappointed that our piglit tests didn't catch this.  I
think we need a new test for [ui]mulExtended that tries more values.  I think
making a big GL_RGBA32I texture that puts the sources in R and G and the
expected results (calculated by C code) in B and A should be easy enough.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20151218/1dc823ab/attachment.html>


More information about the intel-3d-bugs mailing list