scenic
Member
Posts: 126 
Registered on: February 2016
Social Club: Scenic_route
Discord: scenic_route#6404
|
Post by scenic on Feb 27, 2017 20:50:38 GMT
With that last two factors Story's comparison falls flat, as his 60fps run showed less precision on the driver's end and even slid at least once losing a ton of speed. His comparison is greatly skewed in the wrong direction because of large amounts of human error, the comparison is thus invalid. (my bolding above in quote) I understand and appreciate your points, but I strongly disagree with the results being invalid. Inaccurate and imperfect due to driver variation, sure, but still highly relevant as a race pace indicator. Better than any isolated test done in controlled environment. By far! Even I can be fairly consistent around SPA, I know when I do a good lap (no obvious major mistakes), so if I can configure my PC to be fast and slow, I can tell the difference and quantify it simply by doing hotlapping and noting the best times on comparable clean laps. With my current setup, I can do a 3:05 when lucky. If I had a config that would give me boosts on the straight, saving me a second, I would notice immediately. Just did a quick test now, 3 times 4 laps on SPA: 63fps in-game vsync: 3:07, 3:09, 3:08 - best 3:07:79 60fps rivatuner: 3:05, 3:06, 3:05 - best: 3:05:33 140fps rivatuner (110-140): 3:09 (crash), 3:10 (crash), 3:05 - best: 3:05:03 I lose 3-4 seconds by poor driving compared to the best, but I can still see the difference between the settings. Doing more laps, I can quantify quite precisely the advantage. In this case I lose 2,5-3 seconds with Vsync 63fps, and unlocked 140fps (capped rivatuner) has more stutter and sliding, but about same as 60fps. Maybe 125 fps would better, give me CPU some breathing room? Every now and then I'm gonna hit all apexes decently, and that will be enough to tell the difference. I would love to be able to jack up my system to 165 fps and see how much of that 2,5 second that Story got applies to me, if it becomes harder to do the curves at speed and more sliding, maybe I only gain a second, if the benefit comes on the straight, I get the whole 2,5 seconds. Thanks to Story's testing I know approximately how much it benefits him. A drag race on a random section would tell me close to nothing, except maybe hint about things to test on actual track later. You are absolutely right that I cannot explain WHY one is faster, for that isolating the variables and dissecting them in a controlled environment make sense, but I can tell you HOW BIG the difference is, for me, with that car on that track. And that is at the end of the day, the most relevant measure. What vs. Why, two sides of the same thing, neither is "invalid". And of course, direct lap-times might fall short of catching minor differences, the vsync @62/63fps is massive, 2-3%. If there is a 0,5% - 1% difference, it will be harder to tell with a laptime. But an inconclusive test that shows no clear difference is valuable in it self, it is not a test defect, it is a indication of the result being subtle and minor in nature. We pick the low hanging fruits first. One problem with the actual lap-time approach is all the variables to test. Does fullscreen/windowed matter, and does it impact vsync speed? I've got 5 vsync options in my Nvidia CP and two in GTA. What is the graphics setting worth in laptime? Does screen refresh matter? (we can't rule out that it impacts the engine speed, especially if set to be controlled by GTA and not externally, and extra reason for awareness if GTA's own Vsync is enabled, as that directly works of monitor refresh) So maybe 25 different setups to try, doing 4 laps with each would be close to 8 hours, with just one car setup on one track. No thank you. |
|
|
|
Post by SteveJEWBZ on Feb 27, 2017 21:01:03 GMT
Inaccurate and imperfect due to driver variation, sure, but still highly relevant as a race pace indicator. Better than any isolated test done in controlled environment. By far! Pardon me but I do not understand this logic at all.... |
|
scenic
Member
Posts: 126
Registered on: February 2016
Social Club: Scenic_route
Discord: scenic_route#6404
|
Post by scenic on Feb 27, 2017 21:02:32 GMT
Essentially the question we're trying to answer is that of straight line speed, right? No need to try to drive a car around a track with turns to test straight line speed. [...] These findings are easily transferable to the "tracks we actually race", wouldn't you say, scenic ? Easily transferable? You're actually saying that to determine lap-time around a track with X settings, the best way is to run those settings on a straight line and do some extrapolation? Why not just race the god damn track in the first place?!?!?  I'm not trying to answer why fps matters, just how much it matters. |
|
scenic
Member
Posts: 126
Registered on: February 2016
Social Club: Scenic_route
Discord: scenic_route#6404
|
Post by scenic on Feb 27, 2017 21:08:17 GMT
Inaccurate and imperfect due to driver variation, sure, but still highly relevant as a race pace indicator. Better than any isolated test done in controlled environment. By far! Pardon me but I do not understand this logic at all.... The controlled environment is often irrelevant for pace around a track that differs from the controlled environment. I want to know the speed around a track, not the pace on some straight somewhere. The track is right here, a few clicks away. |
|
bladecruiser
Member
Posts: 1,287
Registered on: June 2015
Social Club: BladeCruiser
|
Post by bladecruiser on Feb 27, 2017 22:45:26 GMT
Pardon me but I do not understand this logic at all.... The controlled environment is often irrelevant for pace around a track that differs from the controlled environment. I want to know the speed around a track, not the pace on some straight somewhere. The track is right here, a few clicks away. Then why try and figure out what's happening? It seems like you just want to go look at the hot lap spreadsheet. Check the 60fps lock PC times and compare to the unlocked PC times. Boom, done. |
|
kasumi
Member
wat
Posts: 114
Registered on: August 2015
|
Post by kasumi on Feb 27, 2017 23:05:45 GMT
if Smithy and Kasumi agree no significant difference between (their) 60 and 144, that's pretty solid since they're so good at the game they would notice if there was. maybe end of discussion  please |
|
scenic
Member
Posts: 126
Registered on: February 2016
Social Club: Scenic_route
Discord: scenic_route#6404
|
Post by scenic on Feb 28, 2017 0:21:01 GMT
Then why try and figure out what's happening? It seems like you just want to go look at the hot lap spreadsheet. Check the 60fps lock PC times and compare to the unlocked PC times. Boom, done. The spreadsheet has little if any meaningful data on this topic. We do know that speed/fps effect and cars can change with Rockstar updates, so time of recorded laptime is vital. I guess technically one could compare individual drivers with both 60fps and non-60 time on a track, ask them if it was done at the same time / settings (so it's comparable) and of course what exactly the non-60fps actually was. Aside from Serenniel's quick test, there are almost no reports that 60, 144 and 120-145 fluctuating fps currently differ in any meaningful way, but they did in the past. But it may not be completely equal. Story has found otherwise, running 155-175 fluctuating fps. This is interesting. Story provided the exact data that I find useful, laptimes with different settings, suggesting a glitch in Nero Custom and significant improvement from high FPS for both Osiris and Tempesta. I'm certainly not opposed to understanding what causes the different speeds, but estimating the current actual scope of FPS effect is a more interesting and fundamental question to me. That is best done by us doing our own head-to-head comparisons, to the extent that we can change our settings into undocumented/unknown effect territory. |
|
bladecruiser
Member
Posts: 1,287
Registered on: June 2015
Social Club: BladeCruiser
|
Post by bladecruiser on Feb 28, 2017 1:45:24 GMT
kasumi first, thank you, those two vids are perfect. Then why try and figure out what's happening? It seems like you just want to go look at the hot lap spreadsheet. Check the 60fps lock PC times and compare to the unlocked PC times. Boom, done. The spreadsheet has little if any meaningful data on this topic. We do know that speed/fps effect and cars can change with Rockstar updates, so time of recorded laptime is vital. I guess technically one could compare individual drivers with both 60fps and non-60 time on a track, ask them if it was done at the same time / settings (so it's comparable) and of course what exactly the non-60fps actually was. Aside from Serenniel's quick test, there are almost no reports that 60, 144 and 120-145 fluctuating fps currently differ in any meaningful way, but they did in the past. But it may not be completely equal. Story has found otherwise, running 155-175 fluctuating fps. This is interesting. Story provided the exact data that I find useful, laptimes with different settings, suggesting a glitch in Nero Custom and significant improvement from high FPS for both Osiris and Tempesta. I'm certainly not opposed to understanding what causes the different speeds, but estimating the current actual scope of FPS effect is a more interesting and fundamental question to me. That is best done by us doing our own head-to-head comparisons, to the extent that we can change our settings into undocumented/unknown effect territory. There's a very minor time gap between the two fps settings that could very easily fall within an acceptable margin of error when accounting for car launch on the start of the race or hitting the final checkpoint in the center versus on the side (since the triggers of checkpoints are round and not square). These two videos completely remove any random hidden variables, and also conveniently take racing line out of the equation as well, so you are only left with the results of FPS differences and how bumps may or may not affect them. Perfectly flat surface = no difference at 60fps vs 144fps. Curbs and perfectly flat surface = no difference at 60fps vs 144fps. So that leaves me wondering which variable that got removed is making up the time difference that some people are experiencing. Racing line? Car setup? Time of day? Weather? Vsync (ingame)? Frame lock (gpu instead of ingame)? I would make the assumption that it's most likely slight differences in racing lines that are grabbing more or less curbs is what's accounting for the time differences instead of fps being the culprit, if I'm taking kasumi's videos into account. Granted, there is some evidence that fluctuating fps causes the game engine to freak out, too, so that specifically can't be dismissed either. But thankfully, those videos lend some evidence to stable fps being a non-factor. So, thanks again kasumi.
|
|
kasumi
Member
wat
Posts: 114
Registered on: August 2015
|
Post by kasumi on Feb 28, 2017 2:16:03 GMT
no problem blade. I am playing with locked to 120 fps since a while now btw to just have 100% stable fps
|
|
|
|
Post by SteveJEWBZ on Feb 28, 2017 2:20:56 GMT
But thankfully, those videos lend some evidence to stable fps being a non-factor. This is what I'm setting out to do with the stunt track videos. Isolate the problem. With Kasumi's videos we can see that stable FPS yields essentially the same results, at least for Kasumi. Like I said, I would like to see other people perform the same tests on their machines. If we see different results we can then move on to see what the discrepancy is between the two testing. This is what isolating the problem can do. |
|
kasumi
Member
wat
Posts: 114
Registered on: August 2015
|
Post by kasumi on Feb 28, 2017 2:26:59 GMT
I am playing with santelol since forever and lately I am playing with 120fps lock and he with unlocked @ 144hz and gsync. so his fps fluctuate alot and there is still no difference speed wise. mainly why I lock to 120 to test if I can be even more consistent since the car behavior is not changing 2 much in terms of having different phsysics/traction what so ever because of different fps all the time. with 60 fps I have way more traction. with high fps the car tempts to slide more and with high fluctuating fps my car is unpredictable most of the time and I all of a sudden slide/spin out where I never did before. hard to explain with bad englando and everyone not be able to experience that kind of stuff
|
|
|
|
Post by Sarah on Feb 28, 2017 2:29:42 GMT
what you experience on your system and santelols doesnt matter even the slightest if you have no idea whats causing it, thats the thing we're trying to nail down..
|
|
bladecruiser
Member
Posts: 1,287
Registered on: June 2015
Social Club: BladeCruiser
|
Post by bladecruiser on Feb 28, 2017 2:35:21 GMT
Yea, that's the point of all these discussions - find the culprit behind the voodoo boosts that sometimes happen with FPS differences.
I'm personally hoping to get it figured out so people can break the shit out of the game and get it fixed (very unlikely, but I do try to be optimistic). Or at least post some stupidly crazy lap times.
|
|
|
|
Post by Sarah on Feb 28, 2017 2:51:25 GMT
like, what we really need is for results to be reproducable across systems, and we're nowhere close
|
|
st0cking
Member
Posts: 1,137 
Registered on: November 2015
|
Post by st0cking on Feb 28, 2017 14:26:30 GMT
Not so much related to fps-boosting but a general fps question: I seem to be in 40-60fps range with my vsync on, only tested this in freeroam as the fpscounter seems to crash on me often and i dont need that in a race. So in races i might drop even further than 40 but let's assume i'm always somewhere between 40-60fps, not stable in most locations.
My question is: Would having a stable 60fps improve my times (aside from the fact that it might be easier to drive with smooth/stable fps)? Or does it not matter much that i sometimes drop to as low as 40?
|
|
please
