https://bugs.freedesktop.org/show_bug.cgi?id=99780
Samuel Pitoiset changed:
What|Removed |Added
Resolution|--- |FIXED
Status|NEW
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #12 from Matias N. Goldberg ---
I can confirm applying the patch you uploaded (plus LLVM's patch of course)
fixes the problem.
Thank you.
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #11 from Nicolai Hähnle ---
Created attachment 129675
--> https://bugs.freedesktop.org/attachment.cgi?id=129675&action=edit
Mesa patch
Yes, you need this patch for Mesa as well.
I'm going to pick up the LLVM patch again as well th
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #10 from Matias N. Goldberg ---
I've built LLVM 5.0 from latest SVN.
I've applied the patch from https://reviews.llvm.org/D26348
The issue still persists. I don't know if anything else was needed (e.g.
changes to radeonsi to make use
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #9 from Matias N. Goldberg ---
Thanks for the referral.
I commented there, and they replied back with a potential fix:
> I believe it is very easy to fix, no need to change LLVM: marking the texture
> operation as having unknown si
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #8 from Bas Nieuwenhuizen ---
I think this is the issue described in
https://reviews.llvm.org/D26348
Not idea when the fix is going to go through though.
--
You are receiving this mail because:
You are the assignee for the bug.__
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #7 from Matias N. Goldberg ---
Analyzing the ASM it's clear what's happening: the 3.9.1 version branches
everything.
The 4.0 version branches the calculations, then at the end uses readfirstlane
(this is WRONG) and finally performs on
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #6 from Matias N. Goldberg ---
Created attachment 129536
--> https://bugs.freedesktop.org/attachment.cgi?id=129536&action=edit
ASM from 3.9.1
--
You are receiving this mail because:
You are the assignee for the bug.___
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #5 from Matias N. Goldberg ---
Created attachment 129535
--> https://bugs.freedesktop.org/attachment.cgi?id=129535&action=edit
ASM 4.0
Those readfirstlane instructions are incredibly suspicious
--
You are receiving this mail beca
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #4 from Matias N. Goldberg ---
Created attachment 129534
--> https://bugs.freedesktop.org/attachment.cgi?id=129534&action=edit
IR Generated by LLVM 4.0
--
You are receiving this mail because:
You are the assignee for the bug._
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #3 from Matias N. Goldberg ---
Created attachment 129533
--> https://bugs.freedesktop.org/attachment.cgi?id=129533&action=edit
IR Generated by LLVM 3.9.1
Took me a while to find the IR out of all the noise.
You should consider add
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #2 from Matias N. Goldberg ---
Created attachment 129532
--> https://bugs.freedesktop.org/attachment.cgi?id=129532&action=edit
Vertex Shader used to receive these shadows (already stripped down & with tint)
--
You are receiving th
https://bugs.freedesktop.org/show_bug.cgi?id=99780
--- Comment #1 from Matias N. Goldberg ---
Created attachment 129531
--> https://bugs.freedesktop.org/attachment.cgi?id=129531&action=edit
Pixel Shader used to receive these shadows (already stripped down & with tint)
--
You are receiving thi
https://bugs.freedesktop.org/show_bug.cgi?id=99780
Bug ID: 99780
Summary: Flickering artifacts in radeonsi driver with divergent
texture reads.
Product: Mesa
Version: unspecified
Hardware: x86-64 (AMD64)
OS:
14 matches
Mail list logo
