[Bug 92822] [BDW, bisected] arb_arrays_of_arrays.execution.image_store.basic-imagestore-const-uniform-index

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Sat Dec 12 19:35:40 PST 2015


--- Comment #7 from Timothy Arceri <t_arceri at yahoo.com.au> ---
(In reply to Mark Janes from comment #6)
> The patch seems to have resolved the issue with
> spec.arb_arrays_of_arrays.execution.image_store.basic-imagestore-non-const-
> uniform-index

Thanks for testing Mark.

I have pushed these patches to the piglit repo so the issue with these tests
should be fixed as of feeefb1b47b758.

That makes 2/3 issues being tracked in this bug fixed.

remains to be fixed.

This one requires a Mesa fix, there are two options:

Create a lowering pass the turns all arrays of arrays into seperate single
dimensions arrays.

For example an AoA:  
in my_aoa[2][2][2];

Would become:
in my_aoa[0][0][2];
in my_aoa[0][1][2];
in my_aoa[1][0][2];
in my_aoa[1][1][2];

Where the outer two subscripts are now just part of the variable name rather
than being functional. This will also allow inactive arrays to be easily found.

However it will be a bit of work to get this correct as indirects will need to
be handled somehow as well as things such as assignments of whole AoA.
Also a bunch of clean-up would be needed as its a major change in the way AoA
are currently handled after the initial compile/linking validation is complete.

The other simpler option would be to process the existing IR and store the
values we need for arb_program_interface_query in a newly allocated piece of
memory used just for interface querys.

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/20151213/4c79afcf/attachment.html>

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