GT5's Game-Breaking Online Flaw (OP Updated: 11 Feb)
- Thread starter MGR
- 1,131 comments
- 96,682 views
- 1,019
- johnkiller2
Jon,
you seems to be in a racing group who done the most work trying to figure out what is causing this to happen. The devil might be in details.
I just got fiber yesterday. Ping rate is better and bandwidth, but I cannot get the new submitted router to enable UPnP. I have my next league race on Thursday, and I will use it even if I now has Nat-3 rating.
I by no means state that this is the issue, but it can be something small and/or not logical like this that causes this to happen.
you seems to be in a racing group who done the most work trying to figure out what is causing this to happen. The devil might be in details.
I just got fiber yesterday. Ping rate is better and bandwidth, but I cannot get the new submitted router to enable UPnP. I have my next league race on Thursday, and I will use it even if I now has Nat-3 rating.
I by no means state that this is the issue, but it can be something small and/or not logical like this that causes this to happen.
- 10,456
- United States
- WrecklessAbandon
From what I've seen it's probably a NAT issue. It seems much more common in some NAT 2 and especially NAT 3 users.
NAT 3 is pretty awful itself, I can't find any right this second, but there's a few places 3D3 has gone into detail on how to get better then NAT 3, for some people it'll even get you to NAT 1.
I don't seem to really get this problem, I use a wheel, SDTV/component hookup, have moderately slow internet, NAT 2 connection.
I've had big frame rate drops while racing, to the point where I couldn't see well enough to drive properly, but it doesn't seem to really affect lap times for me.
Edit: Fat PS3
- 3,315
- The Netherlands
- GTP_Sjaak
It happened to me last week,on my fatty.
https://www.gtplanet.net/forum/showthread.php?p=6412927#post6412927
15 racers,all the same car,
Qualification went well but the race had all the bug symptoms.
These are the first lap times of a testrace and the real race of six racers.
1:1'45-1'45
2:1'46-1'48
3:1'47-1'53
4:1'47-2'00
5:1'50-2'01
6:1'52-2'04
Testrace a difference between racer1 and racer6 of 7 sec ,the race 19 sec.
- 4,420
- nealcropper
is there a measurable difference between wheel and ds3? or is it just down to driver skill? and experiance with either?
I'd be very surprised if NAT was the cause, though I'm often proven wrong
I've been through 3 different routers myself, from cheap netgear to high end cisco without a change. Likewise, the boys I race with are all on varied cable / dsl providers, all with different router setups and it seems to pick on anyone 
Perhaps the next logical step would be to try and replicate it in a race. I'm happy to host as I know it's happened while I was hosting before, and I think the SSR5 with an unmodified premium Viper ACRs and Lambo Murchi SVs would be a good start and that's where we all had massive problems.
I can probably get 6 or 7 guys from my normal group to join, so we'd need another 7 or 8 from here to get on the practice pace and fill the room up for the race.
Without wanting to sound like a prick, we'd need a bunch of guys who know exactly what times they can run, to within a few tenths, and do those consistently in practice before racing. Up for it?
- 3,315
- The Netherlands
- GTP_Sjaak
1:1'45-1'45
2:1'46-1'48
3:1'47-1'53
4:1'47-2'00
5:1'50-2'01
6:1'52-2'04
The difference between the nr 1 and nr 6 in the real race was 17 sec.
But when I was timing my own lap(nr5)........i stopped the watch at 1'45 instead of 2'00
The nr 1 racer with 1'45 I clocked at 1'31
2:1'46-1'48
3:1'47-1'53
4:1'47-2'00
5:1'50-2'01
6:1'52-2'04
The difference between the nr 1 and nr 6 in the real race was 17 sec.
But when I was timing my own lap(nr5)........i stopped the watch at 1'45 instead of 2'00
The nr 1 racer with 1'45 I clocked at 1'31
- 10,456
- United States
- WrecklessAbandon
This means it's not only a grip issue, but also a timing issue.1:1'45-1'45
2:1'46-1'48
3:1'47-1'53
4:1'47-2'00
5:1'50-2'01
6:1'52-2'04
The difference between the nr 1 and nr 6 in the real race was 17 sec.
But when I was timing my own lap(nr5)........i stopped the watch at 1'45 instead of 2'00
The nr 1 racer with 1'45 I clocked at 1'31
I can't verify the grip issue typically, but I'll definitely try the timing issue.
nasanu
(Banned)
- 1,429
- Australia
- nasanu
- nasanu
Well you can always find people after an update claiming all sorts of things, for example the people in this very thread telling us how the latest update introduced dirty tyres after going off track (has always been in the game).
This whole thread is worthless until some real testing is done. Timing the replays was one idea but you can not simply test a few 'affected' replays, you must test a large volume of random replays, sort them into groups based on time differences THEN check to see if all the 'affected' replays are all in the same group. That would be some science, checking a handful of replays and finding what you expect to find is just wasting time.
- 1,106
- NC/NoVA
- dudeondacouch
Can anyone upload any of these replays?
I wouldn't mind taking a look at a race where the timing is known to be messed up.
I wouldn't mind taking a look at a race where the timing is known to be messed up.
dsgerbc
Premium
- 2,577
- dsgerbc
I'm confused. You claim that tons of random replays are needed to be checked to prove that the issue exists? I agree that lots of data is needed to isolate what's related to the frequency of it's occurrence. Zero random replays are needed to confirm the existence. Moreover - replays say nothing about existence, it's the driver's claim about a huge discrepancy between quali pace and race pace. If don't want to believe what people are claiming I have a Christmas/NY wish for you: to experience this lag yourself in the coming year. It's quite annoying to start close to the front of the grid and to be seconds a lap slower.
- 1,036
- Vegas
- Trajikone
I would say not worthless as something is happening.
And not limited to just the race, maybe more defined in the race though. Mine happened in Freerun (qualify). WSGTC practice. Everything was off for me by about 1 second. The only reason I noticed was because it was the only car and track I ran for a couple days when I got on. So I knew what my times should be. There were only a couple of racers in the room. Room run by approved stewards. The host was over sea from me. (GTP_Pinoy, I think). When i was in CSL's room my time were normal. CSL you might not remember me telling you at the time, thats why i ran my qualify that day. Cause i knew the time were on. Minus my mistake.
Normal for me was 2:09.2-2:09.6 / 2:10 with mistake
Affected was 2:10.2- 2:10.6 / 2:11 with mistake
ps3 slim/ DFGT/ nat2/ 12+mb/1+mb upload
Last edited:
nasanu
(Banned)
- 1,429
- Australia
- nasanu
- nasanu
So in short... No proof = proof? I can send you links to forums with thousands of people reporting alien abductions, they all can not be wrong can they?
But anyway you misread what I said. Random replays (blindly tested) would be a requirement to prove a connection between replay times and anything else. My interest is in seeing good data gathered and finding out if this is a real effect or just a perception. A bunch of people saying something is only proof that a bunch of people have said something.
In my experience this does not exist but if the data is put in front of me I'd happily change my mind.
And this is where your story can not be taken as fact. I know I sound like a prick here but a persons memory is not a reliable source of data. If you had saved the replays of your practices then you'd have something to go on, as it is you have nothing but a story. Even if you repeated the process that is still a no go because you are not blind and can subconsciously effect the results.
Last edited:
- 4,420
- nealcropper
someone in our leauge had another leauge race yesterday for the fgtc leauge at monte carlo, guess what it was called off because the room was full, 16 drivers, and the guy that is in our leauge is the best driver in our leauge, hes also a ds3 user, and very consistant. He was 7 seconds off quali pace.
- 30
I've been playing online since GT5 came out, doing circuit races usually with the same crowd/rooms, and what I can say is:
- internet connection factors have always been fundamental in terms of lap times/race performance - a higher upload and lower ping to the "host clock" lead to much better lap times and consistent track behaviour
- in general, qualifying and racing lap times have never been perfectly consistent - usually qualifying times were a bit quicker but it could happen frequently having faster racing lap times due to drafting (especially during first laps)
- my online experience has never been consistent - some nights I'm able to keep up with the pack lapping fast and consistently, some other times I'm completely off, were I struggle to control the car, am 2 seconds of pace, bump people by accident in close racing, etc.
- In some tracks consistency and fast lap times are more difficult than in others - to me this has to do with the way the tracks are modeled and how heavy the processing is for the PS3 AND also the track natural features (more or less corners especially), for instance I'm much "slower" in laguna seca, monaco, grand valley than in trial mountain, rome, or spa
- since update 2.02, the number of people in a room has become a much more obvious factor for inconsistencies - I now hear other players in the room complaining much more during races, mainly because of lag during first laps close racing, and the "last night I was lapping 2/3 seconds faster how is this possible" rants are much more frequent now.
I live in Portugal and have a DSL 1Mb max upload connection. I usually play with Brits hosting, using Fiber or DSL connections with similar or faster max upload speeds. To me this as always been the fundamental explanation for the online experience. Players geographically closer and/or with faster connections will be faster (on average, knowing that people do have different levels of skill/experience). Players like me, with much higher latencies will have a harder time online. This has been quite common in online gaming in general and regardless of architecture (dedicated, client/server, p2p, etc) someone will have to hold the master clock and everybody else's events have to somehow be synched. As you guys probably know, some games are already using lag compensation to remove the host and close-to-host advantages.
Where does the advantage come from? Well, the number of client events a player sends to the host clock is higher the better the internet connection. That explains braking distance and "in turn" under/oversteer in GT5 the same way it explains head shot kills in FPS games.
In GT5 the big issue is not a high ping making a bullet reaching it's target much latter than an opponent, it's the synching effort (think of collisions, drafts, etc) making an individual client having to drop a ton of frames in order to keep up with the master. People with better connections will exchange more synching instructions (and in GT5 think of these as grip instructions) with the master hence benefiting from better braking distances and better "in turn control" than people with slower connections. But hey, this is the rule in online gaming .
It's obvious PD changed something in 2.02, which caused bigger game-clock vs stopwatch inconsistencies, and free run. vs race differences, and 16 vs 6 player in a room. But I think whatever happened just worsened things to a point that people who usually had no big online issues are now subject to a degraded experience, to a point most people playing online are feeling pissed off with PD.
And this is really bad news. Unless PD in 2.02 was trying to implement a feature gone wrong (e.g. lag compensation, tyre vs track dynamics) and they now can roll it back, what is going to happen is that things will just stay the way they are right now. I think PD does a far from good job with their testing, many times braking things that used to work, especially in what pertains to the online side of the game (the recent Shuffle Race bug is just inexcusable, you don't need to work in systems development to know what regression testing is).
Lets see what happens with the January update, until then, just race in small rooms, with people geographically close to with, with the best internet connections possible.
- internet connection factors have always been fundamental in terms of lap times/race performance - a higher upload and lower ping to the "host clock" lead to much better lap times and consistent track behaviour
- in general, qualifying and racing lap times have never been perfectly consistent - usually qualifying times were a bit quicker but it could happen frequently having faster racing lap times due to drafting (especially during first laps)
- my online experience has never been consistent - some nights I'm able to keep up with the pack lapping fast and consistently, some other times I'm completely off, were I struggle to control the car, am 2 seconds of pace, bump people by accident in close racing, etc.
- In some tracks consistency and fast lap times are more difficult than in others - to me this has to do with the way the tracks are modeled and how heavy the processing is for the PS3 AND also the track natural features (more or less corners especially), for instance I'm much "slower" in laguna seca, monaco, grand valley than in trial mountain, rome, or spa
- since update 2.02, the number of people in a room has become a much more obvious factor for inconsistencies - I now hear other players in the room complaining much more during races, mainly because of lag during first laps close racing, and the "last night I was lapping 2/3 seconds faster how is this possible" rants are much more frequent now.
I live in Portugal and have a DSL 1Mb max upload connection. I usually play with Brits hosting, using Fiber or DSL connections with similar or faster max upload speeds. To me this as always been the fundamental explanation for the online experience. Players geographically closer and/or with faster connections will be faster (on average, knowing that people do have different levels of skill/experience). Players like me, with much higher latencies will have a harder time online. This has been quite common in online gaming in general and regardless of architecture (dedicated, client/server, p2p, etc) someone will have to hold the master clock and everybody else's events have to somehow be synched. As you guys probably know, some games are already using lag compensation to remove the host and close-to-host advantages.
Where does the advantage come from? Well, the number of client events a player sends to the host clock is higher the better the internet connection. That explains braking distance and "in turn" under/oversteer in GT5 the same way it explains head shot kills in FPS games.
In GT5 the big issue is not a high ping making a bullet reaching it's target much latter than an opponent, it's the synching effort (think of collisions, drafts, etc) making an individual client having to drop a ton of frames in order to keep up with the master. People with better connections will exchange more synching instructions (and in GT5 think of these as grip instructions) with the master hence benefiting from better braking distances and better "in turn control" than people with slower connections. But hey, this is the rule in online gaming .
It's obvious PD changed something in 2.02, which caused bigger game-clock vs stopwatch inconsistencies, and free run. vs race differences, and 16 vs 6 player in a room. But I think whatever happened just worsened things to a point that people who usually had no big online issues are now subject to a degraded experience, to a point most people playing online are feeling pissed off with PD.
And this is really bad news. Unless PD in 2.02 was trying to implement a feature gone wrong (e.g. lag compensation, tyre vs track dynamics) and they now can roll it back, what is going to happen is that things will just stay the way they are right now. I think PD does a far from good job with their testing, many times braking things that used to work, especially in what pertains to the online side of the game (the recent Shuffle Race bug is just inexcusable, you don't need to work in systems development to know what regression testing is).
Lets see what happens with the January update, until then, just race in small rooms, with people geographically close to with, with the best internet connections possible.
- 1,106
- NC/NoVA
- dudeondacouch
That would be a great idea, if only PD would have implemented a filter that actually searches for room characteristics that matter, instead of nonsense like "real" and "fun"
nasanu
(Banned)
- 1,429
- Australia
- nasanu
- nasanu
Has never happened to me, at least not that I can remember. In fact it has been months since I have lost fairly to an equal car.
Then your connection is quite similar to mine, yet I play with many varied connections. I play with other Australians with horrible connections, US and Canadian players with average connections and Japanese and Koreans who have amazing connections. Apart from lag I see no difference at all in general lap times or handling.
No it absolutely does not. In all modern online games you interact with your environment in real time. Each user generates their own game world. You send details of what you did in your game world to the host and receive details of what others did in theirs. Their actions get recreated on your screen. With headshots and FPS games, this is due to the lag in receiving the data of other players movement. Your game world has them in one position, on their screen they had already moved and thus you missed the shot.
In racing games all your car inputs and track feedback are in real time also. Understeer or oversteer can not be explained by lag at all, since the data for this is not leaving your console. You only send and recieve movement data (send yours, receive theirs), that is it. If you are updating that data slower than other players (have a worse connection) then your car will appear jumpy to them as it will constantly correct its path as your console sends out data. It is not at all possible for this to effect the handling of a car.
No... As I just wrote above, this is simply not how it works.
cicua
Premium
- 944
- Copenhagen
- GTP_cicua / cicua
I believe all of this is pure nonsense. You should never have slower lap times because of your connection.
Unlike shooters, each GT5 runs an autonomous race, and just uses incoming data to predict where opponents are. No central arbiter to decide if you got a hit or miss. At least for open lobbies, not sure about private ones. But even in private lobbies, your GT5 alone decides where your car is, and just informs the others about that.
The whole issue here is that individual players' race clock can suddenly run slow, and that must be a problem occuring in that player's GT5/PS3, not the host's.
I wonder if it helps to update the PS3 clock (go to settings and update time via internet). GT5 warns at startup if the clock is off by many seconds, but not if only by a few (why the PS3 does not just autoupdate periodically like any other computer is a different issue). I suspect unsynced clocks could be an issue. If network updates are timestamped.
- 4,420
- nealcropper
cicua
I havn't updated the clock since the clocks went back, didn't think it would affect my lap times, I will update and test then c if it makes a differnce.
- 1,019
- johnkiller2
I'd be very surprised if NAT was the cause, though I'm often proven wrong
Perhaps the next logical step would be to try and replicate it in a race. I'm happy to host as I know it's happened while I was hosting before, and I think the SSR5 with an unmodified premium Viper ACRs and Lambo Murchi SVs would be a good start and that's where we all had massive problems.
I can probably get 6 or 7 guys from my normal group to join, so we'd need another 7 or 8 from here to get on the practice pace and fill the room up for the race.
Without wanting to sound like a prick, we'd need a bunch of guys who know exactly what times they can run, to within a few tenths, and do those consistently in practice before racing. Up for it?
I for sure would be up for an organized test. Hope some other people too as it would be great to pull this off. I can run very consistent laptimes on most tracks.
BTW, the connection I've been using up to now has been Nat 2, average DSL service at 8Mb/s. The problem I have is with my new since yesterday fiber connection where UPnP doesn't work properly hence Nat 3. But I still have my old DSL/Nat 2 connection available.
dsgerbc
Premium
- 2,577
- dsgerbc
No, dude, you'll grasp at straws, challenge statistical methodology and whatever else just not to accept the obvious. Happens every day even to pretty serious scientists.
I've never seen you in any decent online racing series/clubs, so I'm gonna make a hypothesis: you won't be able to tell if you're 4 seconds off your normal consistent pace because you're slow. Go gather some data to prove this wrong.
- 4,420
- nealcropper
it seems logical that if your in game clock is slower in relation to everyone elses then its a fair bet that your game will be to, and if the ps3 or your connection cant hack whats goin on then it may slow a little. Im no computor whizz but it does seem like it could happen?
- 7,207
- Ostend, Belgium
- Speedy6543
Here's a copy (PAL) of the replay of this race.
To be found here on the forum.
I watched the replay again and timed the laps with my watch's stopwatch.
My conclusion here was that the racetimer in GT5 is incorrect.
Here are some compared laptimes all timed on lap 2 of the race.
Myself, GTP_Speedy6543 (DFGT)
I now devided all times in GT5 by the real times to see how long one second is in this GT5 online race.
GTP_Speedy: 113sec/99sec = 1.141414sec
GTP_Gooners: 97sec/85sec = 1.141176sec
GTP_hydro: 111sec/97sec = 1.144329sec
Conclusion: GT5's in race seconds last for ~1.14 real life seconds.
So the timer is equally different for everyone.
However, when you will watch the PAL replay you will notice that the first driver, GTP_Gooners (DS3), is able to maintain higher speeds through the turns than anyone else on track. I think all other drivers where using a wheel. EDIT: GTP_MadMax86 is on DS3 too
So there must be something going on with the grip here.
To be found here on the forum.
I watched the replay again and timed the laps with my watch's stopwatch.
My conclusion here was that the racetimer in GT5 is incorrect.
Here are some compared laptimes all timed on lap 2 of the race.
Myself, GTP_Speedy6543 (DFGT)
- Racetimer: 1'53
- My stopwatch: 1'39
- Racetimer: 1'37
- My stopwatch: 1'25
- Racetimer: 1'51
- My stopwatch: 1'37
I now devided all times in GT5 by the real times to see how long one second is in this GT5 online race.
GTP_Speedy: 113sec/99sec = 1.141414sec
GTP_Gooners: 97sec/85sec = 1.141176sec
GTP_hydro: 111sec/97sec = 1.144329sec
Conclusion: GT5's in race seconds last for ~1.14 real life seconds.
So the timer is equally different for everyone.
However, when you will watch the PAL replay you will notice that the first driver, GTP_Gooners (DS3), is able to maintain higher speeds through the turns than anyone else on track. I think all other drivers where using a wheel. EDIT: GTP_MadMax86 is on DS3 too
So there must be something going on with the grip here.
Last edited:
dsgerbc
Premium
- 2,577
- dsgerbc
We need to procure gooners' replay of that race. Also, have a look at the replay of your N/A race. Just curious how the delta would look like in yours, since you also were some seconds off your quali pace.
Last edited:
- 10,456
- United States
- WrecklessAbandon
Speedy, who was host? (Top of the lounge)
I do think you could benefit greatly from adjusting settings somewhere, no reason at those speeds you should be NAT 3, and NAT 3 has many issues, even if this is not one of them.
It's not internet speed, this I know for certain. It could be NAT type/connection details, or host, I'm not sure.
My internet is slower, around 6-6.5 down/.6 up speed, with no UPnP myself, but NAT 2.
I do think you could benefit greatly from adjusting settings somewhere, no reason at those speeds you should be NAT 3, and NAT 3 has many issues, even if this is not one of them.
- 7,207
- Ostend, Belgium
- Speedy6543
Similar threads
