Just as bad as waiting only for the last fence, after all.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
For the same glmark2 "build" test, perf reports about half
instructions.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Instead of polling the fences regularly, have a thread
that blocks for a single fence using a separate shared
context, then uses eventfd to wake up the main thread
when something happens.
Inside the guest, glmark2 typicially runs twice as fast with the thread
sync. Although in general, the performances seems to be about +30%. The
benefits is mostly for CPU-bounds tasks (when main the thread hits 100%)
A naive perf stat of the vtest renderer with glmark2 "build" test with a
fixed number of frames (500) results in the following stats data:
(do not value timing related informations, since the renderer is ran and
stopped manually)
without thread:
3032.282265 task-clock (msec) # 0.420 CPUs utilized
4,277 context-switches # 0.001 M/sec
102 cpu-migrations # 0.034 K/sec
9,020 page-faults # 0.003 M/sec
7,884,098,254 cycles # 2.600 GHz
4,440,126,451 stalled-cycles-frontend # 56.32% frontend cycles idle
<not supported> stalled-cycles-backend
11,024,091,578 instructions # 1.40 insns per cycle
# 0.40 stalled
# cycles per insn
1,091,831,588 branches # 360.069 M/sec
5,426,846 branch-misses # 0.50% of all branches
with thread:
3403.592921 task-clock (msec) # 0.452 CPUs utilized
7,145 context-switches # 0.002 M/sec
410 cpu-migrations # 0.120 K/sec
6,191 page-faults # 0.002 M/sec
7,475,038,064 cycles # 2.196 GHz
4,487,043,071 stalled-cycles-frontend # 60.03% frontend cycles idle
<not supported> stalled-cycles-backend
9,925,205,494 instructions # 1.33 insns per cycle
# 0.45 stalled
# cycles per insn
834,375,503 branches # 245.146 M/sec
4,919,995 branch-misses # 0.59% of all branches
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
we aren't using this currently bring it back later if required.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
We aren't using this currently, bring it back if we need it later.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Create a bunch of temp arrays based on the input TGSI
ranges. Otherwise the temps code triggers a lack of
some passes in the mesa GLSL compiler.
A large shader from shadertoy triggered 1m30s compile in mesa
due to the large temp array.
Reported-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This updates the tgsi code from mesa, it introduces
changes necessary to deal with tessellation and doubles.
It also drops an unused saturate feature, which we didn't
use anyways.
I'm not sure this is 100% complete, piglit fbo-alpha now passes.
This was a guess as to why gnome-shell was broken, but I don't think
it was the correct one.
- virgl maps VIRGL_FORMAT_B8G8R8X8_UNORM to GL_RGBA8
- _mesa_choose_tex_format() maps GL_RGBA8 to
MESA_FORMAT_R8G8B8A8_UNORM or MESA_FORMAT_A8B8G8R8_UNORM or
MESA_FORMAT_B8G8R8A8_UNORM.
- it ends up in (gl_texture_image *)img->TexFormat
- then in intel (_DRIImage *) image->format.
- on create image, image->dri_format = driGLFormatToImageFormat() (to
ex: MESA_FORMAT_R8G8B8A8_UNORM -> __DRI_IMAGE_FORMAT_ABGR8888)
- on query, intel_lookup_fourcc() -> __DRI_IMAGE_FOURCC_ABGR8888 == GBM_FORMAT_ABGR8888.
For some reason, driGLFormatToImageFormat() doesn't map
MESA_FORMAT_A8B8G8R8_UNORM, so I only check the two possible values I
could find.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
the protocol failed to handle larger shaders, this allow
the renderer to reassemble large shaders and recombined
the chunks before passing them to the GLSL translation.
This also enhances the renderer protocol to allow
for some more info in the shader object, and removes
the separate vs/gs/fs variants in favour of a type field
in the shader.
This code ended up in the other file and really wasn't necessary
there.
Remove the transfer code from virglrenderer.c, move into main
renderer file, and match it with the corresponding transfer
reader.
This should at least fix the crash in compressed textures
with ARB_get_texture_sub_image
This might have some fallout but my HSW doesn't expose this due
to some kernel restrictions, so we should handle the situation
and not call TF2 functions
Dave Airlie
Marc-André Lureau