> Date: Sun, 3 Jun 2001 09:11:49 -0400
>
> bah, this thread is getting boring but the throughput depends on the version
> of avifile you're using and the clip you're using it on. a clip that's
> harder to decode will give you lower results and different versions of
> avifile will give you different results. Also different versions of X
> depending on the acceleration you have available to the V3 will cause
> different results. If anyone wants to compare real performance there needs
> to be some standards set and benchmark developers should have a test clip
> everyone can download.
I already said I was using the same version of avifile 0.53 and the same
sets of clips were used for both (maybe a couple new ones I didn't have
before as well). I also used not only 3.3.6, 4.0.3 and 4.0.99.4 I also
used AcceleratedX in my tests AX had the highest throughput with a 10%
advantage but it was still sort of crap.
The problem was my motherboard, if the bug is fixed in 2.4.5 I can't go
back and confirm it now since I gave the motherboard to my girlfriend for
a windows box. Do you think it possibly isn't a hardware related problem
when I showed an immediate 800% speed increase upon switching only the
CPU+MB? I don't see how this thread even got this far personally I
stopped posting about it after I changed motherboards, I figured the old
MB was just fubar and no one needed to hear my resolution.
I'm not arguing if the v3k is a good card or not, just in my case the
entire problem seems to be due to a kernel driver bug.
And I agree on a set of benchmark standards. I get a 10%+ swing in
throughput rate depending on when I run the test (if I run it 5 times in a
row I get some wild variations sometimes even ) and what the clip is.
a shell script that runs a test clip X number of times and averages it out
or something is maybe what we need, and maybe the script should note the
current load as well?
_______________________________________________
Avifile mailing list
[EMAIL PROTECTED]
http://prak.org/mailman/listinfo/avifile
