Unoptimized builds of multimedia applications

~thong-u/unoptimized/+git/libdrm:ati-4-1-1-branch

  1. Git
  2. lp:~thong-u/unoptimized/+git/libdrm
  3. ati-4-1-1-branch
Last commit made on 2000-12-01
Get this branch:
git clone -b ati-4-1-1-branch https://git.launchpad.net/~thong-u/unoptimized/+git/libdrm

Branch merges

Branch information

Name:
ati-4-1-1-branch
Repository:
lp:~thong-u/unoptimized/+git/libdrm

Recent commits

7950295... by Gareth Hughes <email address hidden>

- Fixed polygon stipple. Was still using SubmitPackets...
- Implemented HW-assisted depth span functions. All access to the depth
    buffer is done via the engine, so we don't have to worry about the
    differences between linear and tiled memory.
- Implemented significantly more 3D-friendly static partitioning of
    offscreen memory. 2D still gets enough pixmap cache to be performant.
    For example, 800x600@16bpp texture memory goes from ~3mb to 11.5mb on a
    16mb card.

aa1a793... by Gareth Hughes <email address hidden>

- Revert texture LRU changes back to 4.0.1 code for now.
- Fix GL_BLEND implementation, correctly fall back to software rendering
    when required.
- Depth buffer tiling for up to 60% performance boost. Need to fix depth
    span functions as a result.
- Fixed scissor. Again. I mean it this time...

1f86294... by Gareth Hughes <email address hidden>

- Fix long-standing scissor bug.
- Fix texture uploads. Now request/upload a single buffer at a time,
    improves available space sharing with other contexts (rather than
    grabbing most of it for a single texture upload). Really need larger
    buffers for better efficiency, this is coming...
- Fix glean makeCurrent failure.
Need to track down some strangeness in the texture LRU mechanism seen with
    Q3 High Quality, but apart from that we're looking in good shape to
    merge with the trunk.

f848f03... by Gareth Hughes <email address hidden>

Fix up merge from trunk.

95c7f54... by Gareth Hughes <email address hidden>

Merged trunk into branch.

edfdc14... by Gareth Hughes <email address hidden>

Merge with latest trunk code. Now requires kernel 2.4.0-test11-pre5 or
    later.

c071d03... by Gareth Hughes <email address hidden>

r128-specific parts of sync with 2.4.0-test11-pre5. Rest is coming shortly.

2a39d83... by Gareth Hughes <email address hidden>

- Cleanup of indirect buffer submission.
- Indexed vertex buffer fixes, more robust handling of indirect buffers for
    the elt path.

81f4398... by Gareth Hughes <email address hidden>

Greatly improve interactivity with MMIO-based 2D acceleration by only
    turning CCE off when the X server actually has to do some 2D rendering.
    We test if the CCE is running at the start of each accel routine, and
    if so we shut it down. If we've shut it down, we start it up again in
    LeaveServer(). An extra if statement in the accel routines is worth it,
    IMNSHO.
Updates to the elt path code. Allow indirect buffer firing to specify start
    and end parameters for buffer reuse. DRM ioctls and wrappers adjusted
    accordingly. Still disabled by default.

79857ad... by Gareth Hughes <email address hidden>

Add support for different primitive types in vertex buffer flush ioctl.
Try and fix Mobility 128 lockups. Seems to lock when shutting down the X
    server from non-standard resolutions, so I've probably messed up the
    CCE ioctls somewhat. Default panel resolution seems to be rock solid...