Ian Romanick wrote:
>
> On Mon, Apr 08, 2002 at 07:03:34PM -0400, Mike A. Harris wrote:
> > On Mon, 8 Apr 2002, Ian Romanick wrote:
> >
> > >I'm sending this as an HTML attachment. If this causes anybody problems,
> > >let me know and I can re-send as plain text.
> >
> > Doesn't cause any problems.. just can't read it. ;o)
>
> D'oh...I guess I should have known better...
>
> All driver combinations were tested with glxinfo, gears, and my Quake3
> tests.
>
> The Quake3 tests involve 6 recorded demos and 5 configurations. All
> Quake3 tests are run at 640x480, with the X server also running at 640x480.
> The configurations are, roughly, as follows:
>
> * fast.cfg - All texture and geometry settings reduced to the lowest
> possible levels.
> * normal.cfg - The "normal" setting for the game.
> * higeom.cfg - Texture settings reduced to the lowest levels, geometry
> settings turned up to the maximum (including 'cg_shadows "3"').
> * texthrash.cfg - Geometry settings reduced to the lowest levels,
> texture settings turned up to the maximum.
> * ultramax.cfg - All texture and geometry settings turned up to the
> maximum.
>
> In the test loop, the configurations are tested in the order listed above.
> All of my recorded demos and configuration files are available upon request.
>
> My test system is a Pentium III 550 with a 16MB G400. The kernel is version
> 2.4.18.
>
> DRM lib MGA Kernel Module DDX/DRI drivers Result
> New New New Pass (1)
> New New Old Pass (1)
> New Old New Pass (2)
> New Old Old Pass (3)
> Old New New Fail (4)
> Old New Old Pass (3)
> Old Old New Fail (4)
> Old Old Old Pass (2)
>
> 1. dribench002.dm_67 hung on the texthrash.cfg setting. The X server
> could be killed, but not restarted. System had to be rebooted. In this
> case, the ultramax.cfg setting was not tested.
> 2. dribench002.dm_67 hung on the texthrash.cfg setting before rendering a
> single frame (i.e., while displaying the "loading" screen). The X
> server could be killed, but not restarted. System had to be rebooted.
> In this case, the ultramax.cfg setting was not tested.
> 3. dribench003.dm_67 hung on the texthrash.cfg setting. The X server
> could be killed, but not restarted. System had to be rebooted. In this
> case, the ultramax.cfg setting was not tested.
> 4. X died during start-up with the message "Fatal server error: Caught
> signal 4. Server aborting"
Ian,
Thanks for the very complete write up.
Based on the results above, I would say the drmCommand changes are okay
for the MGA. Obviously, the MGA driver has some problems that would be
better addressed with the version from the trunk. I'm reluctant to
address the non-drmCommand related bugs on the drmcommand-0-0-1-branch
because there has already been a fair amount of movement on the trunk
since our branch point. For example, Mesa 4.0.2 has been merged in--and
the Radeon TCL and Mach64 drivers could show up on the trunk at any
point. I'd really like to get all drmCommand issues resolved quickly
and get these architectural changes merged in before this branch
diverges further.
Regards,
Jens
-- /\
Jens Owen / \/\ _
[EMAIL PROTECTED] / \ \ \ Steamboat Springs, Colorado
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
Sponsored by http://www.ThinkGeek.com/
