Re: [PATCH xserver 00/19] glamor: Cleanups all over

[Eric Anholt]

Dave Airlie  writes:

> On 28 January 2016 at 18:21, Dave Airlie  wrote:
>> On 28 January 2016 at 11:56, Eric Anholt  wrote:
>>> I've been working on vc4 X performance again, particularly looking at
>>> dragging windows with xcompmgr -c running.  I'll be sending a separate
>>> patch to accelerate a8 rendering on GLES2-class renderers, but even
>>> with that things are crazy slow.  I think at this point the major
>>> problem is the crazy mass of code from rel_sampler().
>>>
>>> This series was some cleanups as I worked toward that, and finally one
>>> patch that should reduce the cost of rel_sampler on non-control-flow
>>> hardware (I see fewer texture sampler requests on vc4).  No
>>> performance change on i965, but I hope that vc4 is in better shape
>>> after it.  (I'm away from the hardware at the moment, but should test
>>> later today).
>>
>> 1-3 look good to me,
>>
>> Reviewed-by: Dave Airlie 
>
> Okay I've dug through the rest, assuming they don't break anything all
> look good.
>
> Reviewed-by: Dave Airlie 
>
> good to know if they have any effect on VC4.

Finally got my giant edifice of scripts working:

8.09199% +/- 0.416675% (n=5) on comppixwin100 on vc4.


signature.asc
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: [PATCH xserver 00/19] glamor: Cleanups all over

[Kenneth Graunke]

On Wednesday, January 27, 2016 5:56:29 PM PST Eric Anholt wrote:
> I've been working on vc4 X performance again, particularly looking at
> dragging windows with xcompmgr -c running.  I'll be sending a separate
> patch to accelerate a8 rendering on GLES2-class renderers, but even
> with that things are crazy slow.  I think at this point the major
> problem is the crazy mass of code from rel_sampler().
> 
> This series was some cleanups as I worked toward that, and finally one
> patch that should reduce the cost of rel_sampler on non-control-flow
> hardware (I see fewer texture sampler requests on vc4).  No
> performance change on i965, but I hope that vc4 is in better shape
> after it.  (I'm away from the hardware at the moment, but should test
> later today).

Series is:
Reviewed-by: Kenneth Graunke 


signature.asc
Description: This is a digitally signed message part.
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: [PATCH xserver 00/19] glamor: Cleanups all over

[Dave Airlie]

On 28 January 2016 at 18:21, Dave Airlie  wrote:
> On 28 January 2016 at 11:56, Eric Anholt  wrote:
>> I've been working on vc4 X performance again, particularly looking at
>> dragging windows with xcompmgr -c running.  I'll be sending a separate
>> patch to accelerate a8 rendering on GLES2-class renderers, but even
>> with that things are crazy slow.  I think at this point the major
>> problem is the crazy mass of code from rel_sampler().
>>
>> This series was some cleanups as I worked toward that, and finally one
>> patch that should reduce the cost of rel_sampler on non-control-flow
>> hardware (I see fewer texture sampler requests on vc4).  No
>> performance change on i965, but I hope that vc4 is in better shape
>> after it.  (I'm away from the hardware at the moment, but should test
>> later today).
>
> 1-3 look good to me,
>
> Reviewed-by: Dave Airlie 

Okay I've dug through the rest, assuming they don't break anything all
look good.

Reviewed-by: Dave Airlie 

good to know if they have any effect on VC4.

Dave.
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: [PATCH xserver 00/19] glamor: Cleanups all over

[Dave Airlie]

On 28 January 2016 at 11:56, Eric Anholt  wrote:
> I've been working on vc4 X performance again, particularly looking at
> dragging windows with xcompmgr -c running.  I'll be sending a separate
> patch to accelerate a8 rendering on GLES2-class renderers, but even
> with that things are crazy slow.  I think at this point the major
> problem is the crazy mass of code from rel_sampler().
>
> This series was some cleanups as I worked toward that, and finally one
> patch that should reduce the cost of rel_sampler on non-control-flow
> hardware (I see fewer texture sampler requests on vc4).  No
> performance change on i965, but I hope that vc4 is in better shape
> after it.  (I'm away from the hardware at the moment, but should test
> later today).

1-3 look good to me,

Reviewed-by: Dave Airlie 

I'll take a better look at the rest tomorrow.

Dave.
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

[PATCH xserver 00/19] glamor: Cleanups all over

[Eric Anholt]

I've been working on vc4 X performance again, particularly looking at
dragging windows with xcompmgr -c running.  I'll be sending a separate
patch to accelerate a8 rendering on GLES2-class renderers, but even
with that things are crazy slow.  I think at this point the major
problem is the crazy mass of code from rel_sampler().

This series was some cleanups as I worked toward that, and finally one
patch that should reduce the cost of rel_sampler on non-control-flow
hardware (I see fewer texture sampler requests on vc4).  No
performance change on i965, but I hope that vc4 is in better shape
after it.  (I'm away from the hardware at the moment, but should test
later today).

Eric Anholt (19):
  ephyr: Make sure we have GLX_ARB_create_context before calling it.
  glamor: Label programs before linking them.
  glamor: Clarify when Render fallbacks happen due to an unsupported op.
  glamor: Drop dead *_from_x_coord_y() functions.
  glamor: Rename the *y_inverted helpers to not say "inverted".
  glamor: Drop comment about dead yInverted flag.
  glamor: Drop dead glamor_pict_format_is_compatible().
  glamor: Set up XV sampler uniforms once at program build time.
  glamor: Simplify XV vertex setup.
  glamor: Convert XV to using glamor_program.c.
  glamor: Drop extra SHADER_IN type for no mask present.
  glamor: Reuse the glamor_program_alpha_* enums for Render.
  glamor: Simplify the pixmap box looping.
  glamor: Drop extra conditionals for large pixmap handling.
  glamor: Clarify some logic in RepeatFix handling.
  glamor: Clean up formatting of RepeatFix shader code.
  glamor: Clarify how the repeat values being passed around work.
  glamor: Cut down a bunch of conditional handling for RepeatFix.
  glamor: Flip around conditionals in RepeatNone fixups.

 glamor/glamor_composite_glyphs.c   |  10 +-
 glamor/glamor_copy.c   |  15 +--
 glamor/glamor_core.c   |  22 ++--
 glamor/glamor_dash.c   |   6 +-
 glamor/glamor_glyphblt.c   |  12 +--
 glamor/glamor_lines.c  |   6 +-
 glamor/glamor_points.c |   7 +-
 glamor/glamor_priv.h   |  35 +++---
 glamor/glamor_rects.c  |   7 +-
 glamor/glamor_render.c | 161 +---
 glamor/glamor_segs.c   |   6 +-
 glamor/glamor_spans.c  |  23 ++--
 glamor/glamor_text.c   |   8 +-
 glamor/glamor_transfer.c   |  16 +--
 glamor/glamor_transform.c  |   9 +-
 glamor/glamor_transform.h  |   3 +-
 glamor/glamor_utils.h  |  70 
 glamor/glamor_xv.c | 213 +
 hw/kdrive/ephyr/ephyr_glamor_glx.c |  34 +++---
 19 files changed, 298 insertions(+), 365 deletions(-)

-- 
2.6.4

___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: glamor cleanups

[Kenneth Graunke]

On Tuesday, March 24, 2015 01:25:49 PM Eric Anholt wrote:
 Here are some documentation fixes and dead code removal from glamor,
 as well as a tiny patch from trying to figure out how some of the
 large pixmap code works.
 
 This can be found on the glamor-cleanups branch of my tree.

Series is:
Reviewed-by: Kenneth Graunke kenn...@whitecape.org


signature.asc
Description: This is a digitally signed message part.
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

glamor cleanups

[Eric Anholt]

Here are some documentation fixes and dead code removal from glamor,
as well as a tiny patch from trying to figure out how some of the
large pixmap code works.

This can be found on the glamor-cleanups branch of my tree.

___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

[PULL] glamor cleanups and VBO changes (and glx fix)

[Eric Anholt]

keithp said he was comfortable with the level of review on these, so
here goes.  I dropped the two changes from the two series that didn't
get review comments: glViewport() in ephyr, and dropping
glXMakeCurrent(None).

The following changes since commit 7f15e5aed79803fa6057fbfda7dfbacacc2e019d:

  Require video drivers to report that they support server managed fds 
(2014-03-10 09:08:46 +0100)

are available in the git repository at:

  git://people.freedesktop.org/~anholt/xserver glamor-pull-request

for you to fetch changes up to 6e5ea5487738cbca9d585d24e66eb65b93ad579c:

  glx: Clear new FBConfig attributes to 0 by default. (2014-03-10 13:31:14 
-0700)


Eric Anholt (14):
  glamor: remove dead global variable.
  glamor: Fix a spelling mistake in GLAMOR_PIXMAP_FBO_NOT_EXACT_SIZE.
  glamor: Add a note about the state of GL_ARB_map_buffer_range.
  dix: Remove incorrect comment about privates.
  glamor: Fix a mismatched glamor_get/put_context().
  glamor: Don't forget to set GL_INVALIDATE_RANGE_BIT on GL_ARB_mbr.
  glamor: Move glamor_emit_composite_vert() to traps, where it's used.
  glamor: Track the next vertex offset as we go in render accel.
  glamor: Track the next vertex offset as we go for non-AA traps.
  glamor: Track the next vertex offset as we go for AA traps.
  glamor: Extract the streamed vertex data code used by Render.
  glamor: Fix requested composite VBO size.
  glamor: Use buffer_storage
  glx: Clear new FBConfig attributes to 0 by default.

 dix/privates.c|   9 +--
 glamor/Makefile.am|   1 +
 glamor/glamor.c   |  17 +
 glamor/glamor.h   |   1 +
 glamor/glamor_gradient.c  |   5 --
 glamor/glamor_pixmap.c|   3 +-
 glamor/glamor_priv.h  |  25 +--
 glamor/glamor_render.c| 108 +--
 glamor/glamor_trapezoid.c | 128 ---
 glamor/glamor_utils.h |   2 +-
 glamor/glamor_vbo.c   | 186 ++
 glx/glxdricommon.c|   2 +-
 12 files changed, 314 insertions(+), 173 deletions(-)
 create mode 100644 glamor/glamor_vbo.c


pgpPDOBx3qzSe.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: [PULL] glamor cleanups and VBO changes (and glx fix)

[Eric Anholt]

Eric Anholt e...@anholt.net writes:

 keithp said he was comfortable with the level of review on these, so
 here goes.  I dropped the two changes from the two series that didn't
 get review comments: glViewport() in ephyr, and dropping
 glXMakeCurrent(None).

 The following changes since commit 7f15e5aed79803fa6057fbfda7dfbacacc2e019d:

   Require video drivers to report that they support server managed fds 
 (2014-03-10 09:08:46 +0100)

 are available in the git repository at:

   git://people.freedesktop.org/~anholt/xserver glamor-pull-request

 for you to fetch changes up to 6e5ea5487738cbca9d585d24e66eb65b93ad579c:

keithp pointed out that he had a problem with the dix comment change,
which I'd missed in processing review.  I've updated to drop the commit,
with a new sha1 for the pull of 96a28e9c914d7ae9b269f73a27b99cbd3c465ac8.


pgpibKNcA07gm.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: [PULL] glamor cleanups and VBO changes (and glx fix)

[Keith Packard]

Eric Anholt e...@anholt.net writes:

 Eric Anholt (14):
   glamor: remove dead global variable.
   glamor: Fix a spelling mistake in GLAMOR_PIXMAP_FBO_NOT_EXACT_SIZE.
   glamor: Add a note about the state of GL_ARB_map_buffer_range.

   dix: Remove incorrect comment about privates.

   glamor: Fix a mismatched glamor_get/put_context().
   glamor: Don't forget to set GL_INVALIDATE_RANGE_BIT on GL_ARB_mbr.
   glamor: Move glamor_emit_composite_vert() to traps, where it's used.
   glamor: Track the next vertex offset as we go in render accel.
   glamor: Track the next vertex offset as we go for non-AA traps.
   glamor: Track the next vertex offset as we go for AA traps.
   glamor: Extract the streamed vertex data code used by Render.
   glamor: Fix requested composite VBO size.
   glamor: Use buffer_storage
   glx: Clear new FBConfig attributes to 0 by default.

Merged (except for the incorrect dix change)
   7f15e5a..96a28e9  master - master

-- 
keith.pack...@intel.com


pgpOh7cN8IHII.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

little glamor cleanups (and a dix cleanup)

[Eric Anholt]

Here are some trivial patches from the glamor-server branch for
review.  I'm planning to send out the VBO changes next, which get us
closer to making glamor not actually suck for performance.

These are on the glamor-cleanups branch of my tree, branched from
glamor-pull-request.

___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Re: little glamor cleanups (and a dix cleanup)

[Markus Wick]

Patches 1-5 are:

Reviewed-by: Markus Wick markus at selfnet.de

Am 2014-03-06 18:00, schrieb Eric Anholt:

Here are some trivial patches from the glamor-server branch for
review.  I'm planning to send out the VBO changes next, which get us
closer to making glamor not actually suck for performance.

These are on the glamor-cleanups branch of my tree, branched from
glamor-pull-request.

___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel