[Mesa-dev] r600 sb tessellation support
Gert Wollny
gw.fossdev at gmail.com
Wed Jan 10 10:10:14 UTC 2018
Am Mittwoch, den 10.01.2018, 16:48 +1000 schrieb Dave Airlie:
> This is an attempt to add tessellation support to the SB backend.
>
> The main things needed are GDS access which is used for tess
> factor storage (also used for atomic counters), and LDS access
> which is needed to pass all the data between stages.
>
> The first 19 patches are the stuff I'm happy with, the
> nop/sanity shader tests pass with those (and sb enabled).
>
> The last two patches make heaven work and turn on sb,
> I'm not suggested these be applied as-is yet.
Compared to yesterdays 42 regressions nosb versus sb it's now down to
five regressions within -t tessellation:
executionÂ
1in-1outÂ
tes-input/tes-input-gl_clipdistance
tes-input/tes-input-patch-mat2x4_2
tes-input/tes-input-patch-mat3x4_2
tes-input/tes-input-patch-mat4
and one failure became a crash in
trivial-tess-gs_no-gs-inputs
but the crash can easily be alleviated with a patch I send out later.
(I tried to send it yesterday, but noted today that something had gone
wrong, i.e. I did send an empty email instead).
Some numbers on 6870HD:
Heaven 1280x1024
Quality: High, Tesselation: Normal, Anti-Aliasing: Off
FPS: 18.5 [5.7, 70] (was: 12.6 [4.0, 61] before any optimization).
Tessmark x32, 1024x640
FPS: 45, 2153 points (was 10, 635)
With all this: Tested-By: Gert Wollny <gw.fossdev at gmail.com>
> I think in theory enabling sb for atomics/images/compute should
> be fine after this series as well, but I haven't tested that too
> much.
I'll check this out.
Many thanks for your work on this,
Gert
More information about the mesa-dev
mailing list