Xv, radeon-exa-16.diff and some observations

[Dennis Jacobfeuerborn]

Eric Anholt wrote:
> On Mon, 2005-09-12 at 22:16 +0200, Dennis Jacobfeuerborn wrote:
>> I have tested the Xorg tree from last night right after the radeon EXA 
>> patch went in and things are working ok so far on my 9500pro card.
>> The EXA enabled driver version seems to be slower though than the one in 
>> 6.8.2. What is the best way to benchmark this? Scrolling in Firefox for 
>> example is visibly slower and more "jerky" than in the 6.8.2 release which 
>> surprised me a bit (The logfile says that the driver is "Using EXA 
>> acceleration architecture").
> 
> Do you see any non-antialiased fonts on your window?  It looks like this
> is what's killing us there, from my testing (and benh had previously
> reported this for the case of his terminal).  I'm not sure how we want
> to deal with this quite yet.

Not sure if this benchmark counts as scientific but I executed the 
following command in xorg cvs and xorg 6.8.2 in both gnome-terminal 
(antialiased) and xterm (non-antialiased):

time for((i=0;i<100;i=i+1));do cat Xorg.0.log ;done


Here are the results:

gnome-terminal xorg 6.8.2:
real    0m23.495s
user    0m0.044s
sys     0m0.448s

xterm xorg 6.8.2:
real    0m10.021s
user    0m0.024s
sys     0m0.452s

gnome-terminal xorg cvs:
real    0m34.595s
user    0m0.044s
sys     0m0.220s

xterm xorg cvs:
real    0m24.031s
user    0m0.056s
sys     0m0.244s

Regards,
   Dennis