RenZO
02-21-2009, 07:51 AM
Hello,
I made about 300 gaming tests with pts (10 games, 3 resolutions, 10 cards). To do this, I had to change settings in order to run each test only once. Because with games, running 3 times is not usefull, results are always identical.
Another hint: With etqw-demo, first test is always higher, because running in 640x480. Reverse problem with doom3 and quake4, first test is always lower (and usecache isn't different). So if we do an average on 3 tests, it's false.
With pts 1.6, I use to do, for example:
phoronix-test-suite benchmark doom3 doom3 doom3
And I choose different resolutions for each test, I have 3 tests like i want. So I think changing "Times to run" to 1 is a good idea, for games.
Hope my feedback could help, thanks :)
EDIT: Maybe "Times to run" could be a variable?
I made about 300 gaming tests with pts (10 games, 3 resolutions, 10 cards). To do this, I had to change settings in order to run each test only once. Because with games, running 3 times is not usefull, results are always identical.
Another hint: With etqw-demo, first test is always higher, because running in 640x480. Reverse problem with doom3 and quake4, first test is always lower (and usecache isn't different). So if we do an average on 3 tests, it's false.
With pts 1.6, I use to do, for example:
phoronix-test-suite benchmark doom3 doom3 doom3
And I choose different resolutions for each test, I have 3 tests like i want. So I think changing "Times to run" to 1 is a good idea, for games.
Hope my feedback could help, thanks :)
EDIT: Maybe "Times to run" could be a variable?