Hi guys,
Edit: I'm only going to post my own findings not others as it's not fair if they don't want to be drawn into it...
Right so I've heard this circulating for a little while, sorry for the grandiose title but I think it's important for leagues on here to know what is going on. I spent my time on Saturday after release completing scenarios from which I had the top 10 scores for 15/20 of them. Fantastic I thought, I'm sorted for on-line already. Then Saturday evening during a pre-season league race I noticed I would use full KERS and DRS in Nurburgring (not the highest speed of tracks) and watch people *still* pull away from me. Sunday, 5 lap races and noticed I averaged about 1.5 seconds a lap slower than a couple of people, 2.7 seconds around one track...I managed only 2 victories after around 20 races compared to my 48% victory rating in 2012. Tonight I managed 6 out of 6 victories...
Many wont know me... I'm a twitchy paranoid driver who ultimately isn't as fast as some but I'm ridiculously consistent hence the concern I had; I thought people were hiding set-ups but they just reported they were using 2012 set-ups (sorry to those whom I doubted for a brief time).
I tried numerous hosts with the same people, tried lowering settings, extreme set-ups, all to no avail. Then I stopped recording and noticed a small reduction in the gap but reported it to people as inconclusive after complaining to them all night, noting that a couple of tenths wasn't enough to draw conclusions... Today I get home from work and Shaki and Starplayer both informed me about the FPS bug.
The basic premise is, the higher your FPS, the faster you go. This is the opposite of the rumours circulating around 2012. Of course in 2012 it wasn't feasible that you could run at 20 fps to test the theory and still be accurate but now it's reversed you can all test it. Go into on-line, make the conditions stable (sunny) and use the same set-up with both minimum and maximum graphics settings. This will not work offline; logic dictates that if all cars are controlled by your computer (offline) you'll all get the same reduction, probably culminating in lap times which do not sync to real life but do sync to realistic lap times against AI... it has to be on-line as sync issues, latency and differing computer set-ups all occur there, increasing the likelihood of issues. Hence, scenario and other such modes will not be affected by this... 99% certain.
So my final test, two 5 lap races around Spa, the first of which was recorded at 140FPS, the second at 50FPS. If anyone tried to overtake, I backed off so I could get clean air. My fastest lap with 140FPS was recorded with medium fuel, 50FPS on high fuel.
140FPS Medium Fuel:
50FPS High Fuel:
Oh and for good measure, the results when not everyone knows about this issue... I felt horrible after this race but please remember, I was a long long way off the pace prior to trying this. Sorry there aren't more but I don't want to win races like this
I'd like to propose that F1 2013 PC Leagues don't start until this is sorted although I'm pretty certain you're all already aware of this. I realise this may make others do the same but I don't want people to feel the utter frustration that I did whilst racing not knowing what was going on...
Also please note that whilst at 140FPS, my steering wheel would basically attempt to steer me around corners. The G27 isn't the strongest of wheels but when it attempts to full lock in one direction, seemingly with no notice, it is extremely bizarre to drive. Going from 7th gear down to 2nd around a right hander with say an average braking point of 120metres, at about 90metres, the wheel would pull at 100% force to the right during the braking zone which you would have to correct, hence the fact that I think I could have gone faster with 140FPS if I had of turned the FFB to a lower setting or gotten used to the weird effect. At some points it turns off altogether (Monaco anywhere below 3rd gear) at some places it tries to veer you off the track (Brazil, last proper corner). Not sure if this has any impact on where this coding fault lies but worthy of looking at anyway.
I hypothesise that this is based on someone trying to improve performance for console versions where a change needed to be made for the PC port. On consoles, everyone has roughly the same FPS so no problem... might I suggest that if anyone happens to read this from CM, look at your timing counter and look for where someone has rounded up or down a number to improve performance. Maybe an implementation of a customised .ceil() or .floor() which would affect rounding numbers dependant on the number of FPS per second... any float truncation? It's going to be in your on-line code base...
Codemasters, you've made a game which in certain parts is particularly epic; if this is all resolved you will have a winner here, please don't let it go to waste... timing bugs in a racing game, especially one which tries on hundredths of a second to win or loose cannot remain.
Kind Regards,
Mark.
Edit: I'm only going to post my own findings not others as it's not fair if they don't want to be drawn into it...
Right so I've heard this circulating for a little while, sorry for the grandiose title but I think it's important for leagues on here to know what is going on. I spent my time on Saturday after release completing scenarios from which I had the top 10 scores for 15/20 of them. Fantastic I thought, I'm sorted for on-line already. Then Saturday evening during a pre-season league race I noticed I would use full KERS and DRS in Nurburgring (not the highest speed of tracks) and watch people *still* pull away from me. Sunday, 5 lap races and noticed I averaged about 1.5 seconds a lap slower than a couple of people, 2.7 seconds around one track...I managed only 2 victories after around 20 races compared to my 48% victory rating in 2012. Tonight I managed 6 out of 6 victories...
Many wont know me... I'm a twitchy paranoid driver who ultimately isn't as fast as some but I'm ridiculously consistent hence the concern I had; I thought people were hiding set-ups but they just reported they were using 2012 set-ups (sorry to those whom I doubted for a brief time).
I tried numerous hosts with the same people, tried lowering settings, extreme set-ups, all to no avail. Then I stopped recording and noticed a small reduction in the gap but reported it to people as inconclusive after complaining to them all night, noting that a couple of tenths wasn't enough to draw conclusions... Today I get home from work and Shaki and Starplayer both informed me about the FPS bug.
The basic premise is, the higher your FPS, the faster you go. This is the opposite of the rumours circulating around 2012. Of course in 2012 it wasn't feasible that you could run at 20 fps to test the theory and still be accurate but now it's reversed you can all test it. Go into on-line, make the conditions stable (sunny) and use the same set-up with both minimum and maximum graphics settings. This will not work offline; logic dictates that if all cars are controlled by your computer (offline) you'll all get the same reduction, probably culminating in lap times which do not sync to real life but do sync to realistic lap times against AI... it has to be on-line as sync issues, latency and differing computer set-ups all occur there, increasing the likelihood of issues. Hence, scenario and other such modes will not be affected by this... 99% certain.
So my final test, two 5 lap races around Spa, the first of which was recorded at 140FPS, the second at 50FPS. If anyone tried to overtake, I backed off so I could get clean air. My fastest lap with 140FPS was recorded with medium fuel, 50FPS on high fuel.
140FPS Medium Fuel:
50FPS High Fuel:
Oh and for good measure, the results when not everyone knows about this issue... I felt horrible after this race but please remember, I was a long long way off the pace prior to trying this. Sorry there aren't more but I don't want to win races like this
I'd like to propose that F1 2013 PC Leagues don't start until this is sorted although I'm pretty certain you're all already aware of this. I realise this may make others do the same but I don't want people to feel the utter frustration that I did whilst racing not knowing what was going on...
Also please note that whilst at 140FPS, my steering wheel would basically attempt to steer me around corners. The G27 isn't the strongest of wheels but when it attempts to full lock in one direction, seemingly with no notice, it is extremely bizarre to drive. Going from 7th gear down to 2nd around a right hander with say an average braking point of 120metres, at about 90metres, the wheel would pull at 100% force to the right during the braking zone which you would have to correct, hence the fact that I think I could have gone faster with 140FPS if I had of turned the FFB to a lower setting or gotten used to the weird effect. At some points it turns off altogether (Monaco anywhere below 3rd gear) at some places it tries to veer you off the track (Brazil, last proper corner). Not sure if this has any impact on where this coding fault lies but worthy of looking at anyway.
I hypothesise that this is based on someone trying to improve performance for console versions where a change needed to be made for the PC port. On consoles, everyone has roughly the same FPS so no problem... might I suggest that if anyone happens to read this from CM, look at your timing counter and look for where someone has rounded up or down a number to improve performance. Maybe an implementation of a customised .ceil() or .floor() which would affect rounding numbers dependant on the number of FPS per second... any float truncation? It's going to be in your on-line code base...
Codemasters, you've made a game which in certain parts is particularly epic; if this is all resolved you will have a winner here, please don't let it go to waste... timing bugs in a racing game, especially one which tries on hundredths of a second to win or loose cannot remain.
Kind Regards,
Mark.
