[Mesa-dev] [PATCH] linker: Remove erroneous multiply by 4 in uniform usage calculation

[Kai Wasserbäch]

Dear Ian,
Ian Romanick schrieb am 21.11.2011 20:51:
> From: Ian Romanick <ian.d.romanick at intel.com>
> 
> The old count_uniform_size::num_shader_uniforms was actually
> calculating the number of components used.  Multiplying by 4 when
> setting gl_shader::num_uniform_components caused us to count 4x as
> many uniform components as were actually used.
> 
> Signed-off-by: Ian Romanick <ian.d.romanick at intel.com>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42930
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42966
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43125
> Cc: Vinson Lee <vlee at vmware.com>
> Cc: Marek Olšák <maraeo at gmail.com>
> ---
>  src/glsl/link_uniforms.cpp |   11 ++++++-----
>  1 files changed, 6 insertions(+), 5 deletions(-)
> [...]

I can give you a
Tested-by: Kai Wasserbäch <kai at dev.carbon-project.org>
on this patch. The reason why I bothered with investigating this myself further
was, that I had initially seen a regression for glsl-fs-uniform-array-6 between
git/9ae10e9c and git/6246c217 (with r600g on a R700 class chip; see [0]) and my
initial bisecting pointed to 8e366dc365d01213b71b87ace47d30938db74845. But
glsl-fs-uniform-array-6 kept failing after Vadim's fix for 8e366dc3, so I wanted
to make sure this is not something left over from 8e366dc3 and a new Piglit run
with your patch applied on top of git/03df791c lets glsl-fs-uniform-array-6 pass
again with r600g on my GPU.

Kind regards,
Kai Wasserbäch


[0] <http://dev.carbon-project.org/misc/piglit/regressions.html>



-- 

Kai Wasserbäch (Kai Wasserbaech)

E-Mail: kai at dev.carbon-project.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 665 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20111122/c637ea18/attachment.pgp>