Well, here's how I'd benchmark a card:
To hell with framerates. I'll be measuring frame intervals.
Basically I'd rig up my tests to record say, the 5 longest intervals between frames and average them. Of course, that would be out of a run of a fair number of minutes & repetitions and the first run-through would be ignored. Also, tests would be run twice- once w/ vsync on, once with it off.
Scores would be measured in milliseconds, lower is better.
Basically it's a test of how much your card can suck, not how good it can be.
To hell with framerates. I'll be measuring frame intervals.
Basically I'd rig up my tests to record say, the 5 longest intervals between frames and average them. Of course, that would be out of a run of a fair number of minutes & repetitions and the first run-through would be ignored. Also, tests would be run twice- once w/ vsync on, once with it off.
Scores would be measured in milliseconds, lower is better.
Basically it's a test of how much your card can suck, not how good it can be.
Comment