Created attachment 129692<=
/a> [details]
carrizo run with HEAD
Tessellation is broken on Carrizo (A12-9800) hardware and apparently has be=
en
for a while. The offending commit is=20
commit a4e2146a9d24592ed7e3bf778e3c21c6cfb89330
Author: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Date: Mon May 2 14:55:52 2016 +0200
radeonsi: Use buffer loads and stores for passing data from TCS to TES.
We always try to use 4-component loads, as LLVM does not combine loads
and they bypass the L1 cache.
We can't use a similar strategy for stores and this is especially
notable with the tess factors, as they are often set with separate
MOV's per component in the TGSI.
We keep storing to LDS and the LDS space, so we can load the outputs
later, either due to the shader, of for wrting the tess factors.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai H=C3=A4hnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Ol=C5=A1=C3=A1k <marek.olsak@amd.com>
I found it via running Unigine-Heaven with tessellation enabled.
The bug doesn't occur on other VI hardware (Polaris10, FIJI).
I can confirm that the HEAD~ commit leads to spec/arb_tessellation_shader/*
tests (in piglit) passing compared to running with HEAD.