If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
EDIT: Actually I got it backwards. I said frametimes were calculated from framerates, but that's not it, framerates are calculated from frametimes. My bad.
Making a reasonably generic LD_PRELOADable library that would hook into generic function (glSwapBuffers, ...) and measure frames per second shouldn't be a unsolvable problem (this is how FRAPS does it on windows basically).
One possible downside is that PTS would have to require things to run in insecure mode (assuming it doesn't already, truth be told, I never checked), so VAC (or some other anticheat) won't trigger since cheats generally hook this function and derive GL context and objects via clever hacks from there.
It is not exactly trivial to automate a benchmark for a game that does not provide a reliable, standardized benchmarking scenario where every run is nearly identical. Replaying a recorded macro of a user input does not work as any timing related changes fully mess up the outcome.
Actually sending commands, i.e press tilde, write in command, press enter, etc could work pretty well by just waiting more-than-enough time after each step.
I am not sure it would work for the gameplay (=benchmark) itself. For some games it should work nicely, but I suspect there exist games which would require image capturing and comparison in order to determine which key, mouse button or keypad button to send to the game.
In the beginning, we should ignore the complex cases (=image analysis) and hope that it will be possible to implement the required algorithm later when the need comes.
What about like a input capture/replay mechanism? Something where you would have to manually record keyboard/mouse/gamepad input one time and then replay it to reproduce the exact same scenario?
I'm just asking because I don't know if such a thing exists or what the technical limitations of it would be.
I am not sure it would work for the gameplay (=benchmark) itself. For some games it should work nicely, but I suspect there exist games which would require image capturing and comparison in order to determine which key, mouse button or keypad button to send to the game.
In the beginning, we should ignore the complex cases (=image analysis) and hope that it will be possible to implement the required algorithm later when the need comes.
I meant for simply starting the benchmark, if there is no command line flag.
Recording and playbacking gameplay itself would probably be pretty complicated.
Btw, some games allow to record gameplay (i.e rts games, gta-s) and playbacking the recording could also be used for benchmarking - but there are not much games which allow that.
Comment