Very important question about FPS

General discussion about the games and the scene.

Moderator: Moporators

Post Reply
User avatar
jaytea
37mins club
Posts: 994
Joined: 22 May 2004, 08:45

Very important question about FPS

Post by jaytea » 10 Sep 2019, 06:57

Has anybody ever confirmed that FPS is the one and only universal indicator of Elma performance? That Elma running at X fps on one PC is expected to behave exactly the same as Elma running at X fps on another PC with regards to probability and frequency of wheel pops, bounces, hangs, bugs, etc.? That is to say, regardless of any other differences between the two PC environments (whether one is a PC and one a laptop; whether one is running QuickTime, YouTube, whatever).

Could somebody please confirm this once and for all? I missed the memo.
im pretty good at elma

User avatar
Grace
38mins club
Posts: 4710
Joined: 19 Nov 2005, 10:45
Location: Deep in your Imagination, Twirling your Dreams and Weaving your thoughts.

Re: Very important question about FPS

Post by Grace » 10 Sep 2019, 09:48

Short answer: no, fps isn't the only thing

Medium answer: There have been some interesting anomalies observed with weird, seemingly unrelated stuff happening on the pc. If you have a youtube video running (whether it's on a second screen, same screen w/ windowed or not displayed whatsoever) the bike behaves quite differently than without a video running, regardless of if fps isn't changed. IIRC loop-de-loop for example is slightly easier to get a good time with a yt vid running. I believe pawq has done some tests of this with audio playing too and I vaguely remembering people deciding it was specifically the sound causing the difference, as little sense as it makes. I'm sure there's more things like this too.

Long answer: someone else with more elma science knowledge may be able to elaborate better than I can.
Image Cyberscore! Image
___________________________________________________
Image
Targets: 6 Legendary, 18 WC, 25 Pro, 5 Good | 37 Australian Records | AvgTT: 40:11:92

User avatar
pawq
38mins club
Posts: 6138
Joined: 24 Aug 2008, 19:56
Team: TR
Location: Southampton, UK

Re: Very important question about FPS

Post by pawq » 10 Sep 2019, 11:39

Grace wrote:
10 Sep 2019, 09:48
I believe pawq has done some tests of this with audio playing too
Yeah, I noticed big variations depending on whether audio was playing or nat. It was specifically about front wheel lifting off the ground when throttling on a flat surface. No audio made the Enigma start impossible :facepalm: I then had it again in some bullet battle at FEM 19, but that wasn't about audio, just the wheel lifted up enough only about 1 out of 5 times, with only gas pressed at start.

But, correct me if I'm wrong, just the fact that you have unlimited fps (and shows as >1000 in counters), or limited to a specific value, doesn't necessarily mean that it's actually constant, and that it doesn't fluctuate slightly below those limits. Also, the main premise of okeol was setting const fps of physics calculations so that elma will behave identically on any pc.

tl;dr: should probably wait for mila or jon to answer

User avatar
jonsykkel
Kuski
Posts: 910
Joined: 24 Nov 2009, 20:53
Team: BAP
Contact:

Re: Very important question about FPS

Post by jonsykkel » 10 Sep 2019, 17:55

physics timestep is calculated based on numbers from winapi func timeGetTime() which returns integer # of ms since system start. thats the only thing that affects physics
timeGetTime() gets numbers from sum clock that is updated at a certain frequency which is a global setting that can be changed by any process running on the system (the actual frequency used will corespond to the highest resolution requested by any process)

at some point someone has patched elma exe so it requests 2ms resolution
but bakground proceses wil most likely request a higher resolution so the resolution used wil be <2ms in practice (on my pc it was set to exactly 1.25ms at first when i started gomputer, 1sec/0.00125 = 800fps btv which explains sum cumputers geting 800fps physics even if graphics framerate is much higher)
Image
Prestigious member of 16.6x IL101 club

User avatar
pawq
38mins club
Posts: 6138
Joined: 24 Aug 2008, 19:56
Team: TR
Location: Southampton, UK

Re: Very important question about FPS

Post by pawq » 10 Sep 2019, 20:41

Wow, thanks a lot for the explanation!

Maybe silly question, but would it in theory be possible to manually request a lower-than-anything-else timestep so that it (hopefully) remains constant? If processes can then I guess user should too? And how would one go about checking this on their own pc?

User avatar
jonsykkel
Kuski
Posts: 910
Joined: 24 Nov 2009, 20:53
Team: BAP
Contact:

Re: Very important question about FPS

Post by jonsykkel » 10 Sep 2019, 21:05

pawq wrote:
10 Sep 2019, 20:41
Wow, thanks a lot for the explanation!

Maybe silly question, but would it in theory be possible to manually request a lower-than-anything-else timestep so that it (hopefully) remains constant? If processes can then I guess user should too? And how would one go about checking this on their own pc?
yes, you can run this virus i made, it tells you wat the current resolution is and then sets it to 1ms
the os seems to reset the frequency when the process exits so has to be running in bakground for it to vork
it could be fixed in the eol exe also if it wasnt for anti xiit disconecting you if you modify the exe

http://zzz.ink/elma/tools/1000fps.exe
http://zzz.ink/elma/tools/1000fpsh.exe (hidden window, go ctrl alt delete to stop)
Image
Prestigious member of 16.6x IL101 club

User avatar
Zweq
35mins club
Posts: 3933
Joined: 28 Nov 2002, 15:54
Location: suo mesta

Re: Very important question about FPS

Post by Zweq » 11 Sep 2019, 06:33

yy youtube video on background made brutals in hi flyer smoother, proved it in milasl by loading at the brutal spot where i didn't have to time any input. We ar talking about difference of making 25.xx or not making the brutal at all. Why it hapens? it probably affects frame timesteps somhow, making them either more consistent or less consistent, or in other words less variance or more variance. I have no idea which one is better for smooth brutals. I guess it would be easy to test in okesl is completely static fps the best for smooth brutals.
Image

Post Reply