[ut2004] UT2004 benchmarking: what does the "rand" value mean?
Corey Hickey
bugfood-ml at fatooh.org
Wed Apr 21 20:31:40 EDT 2004
Xipher wrote:
> is their any correlation between the rand number, and the average FPS of
> the benchmark?
>
Yes, actually, there is. The difference is slight, but definitely seems
to be there. I was going to mention that, but wanted to get a larger
sample size first. Here's what I have so far:
rand frequency avg. of avg. fps std. deviation
1425745073 2 63.983518 0.451851
1919451183 28 64.612598 0.278567
2051811764 9 64.534478 0.207600
560406543 11 64.205582 0.225971
Obviously, the data for the first row is shaky because there aren't
enough samples. The difference between the second row and the forth row
looks significant, though.
I'll run the tests overnight and see what happens. Incidentally, bug
#1432 (ut2004 never quits) makes automated benchmarking a bit annoying.
I've had to do it like this:
-----------------
for (( i=1 ; i <= $num ; ++i )) ; do
utbenchmark &
sleep 50
killall -SIGINT ut2004-bin
done
-----------------
If somebody know a better way, please tell me. :)
Correction:
I mentioned in the last email that it was a one-minute test. I was mixed
up, actually. The ut2004 command has -seconds=77, and it runs the game
for about 36 seconds (excluding loading time).
-Corey
More information about the ut2004
mailing list