GT5's Game-Breaking Online Flaw (OP Updated: 11 Feb)

[Spagetti69]

Can your cars roll over online?
I seem to remember that couldn't.
Or maybe I'm imagining it.

 

[dudeondacouch]

No, they cannot.

 

[ErikAndre71]

We are trying to determine specifically what circumstances cause the issue and how to avoid it. Once we've properly analysed the issue, we'll be posting about it. The help the GTP community can provide is not asking why we're "hesitant" to post it on the blog (we aren't) or when we're going to (when we have sufficient information) but by providing more and more relevant data.

The only way around it is to limit the lobby to 10 people. We are several people not bothering to post here because _all_ the sufficient information is already here.

The Gran Tursimo box say its capable of a 16 players online experience, it's not. It's not the owner of this forum or the users of this forum to figure out why. PD are the only one who could fix this.
We could sit down and speculate for months or even years so if you don't hesitate to post. Well it's about time.

 

[dsgerbc]

I'm confused. Either PD is already aware of the issue (like some important members seemed to imply) and then we don't really have to gather anything, or they aren't. In which case all we really should have to do - run a bunch of races with >13 players repeatedly and save the replay.

 

[OwensRacing]

This really has got some impressive attention.

The ole gut says no worries. It's being handled. . I truly think this site is watched by the big players more then we give credit. Jordan seems to have gained respect by many in the industry and around GT5. So that says a lot in itself.

 

[FlyinHawaiian77]

Last night I was trying to do a league race. We started a couple of one lap races to see if we would come into any problems. Those races were fine for me. Had a black screen for a little while and the race started.

So we finally get to the point to where we are starting the real race. It was 35 laps around Sarthe. 1st try at starting the race, black screen. Had to reboot the ps3. Second time around, the host starts the race and I get the usual countdown. This time after the clock reaches 0, another countdown starts but it starts from 7. After that countdown went to 0, the dialog box for voting for a track comes up. Says "voting will close in" and then it was blank for the timer.

While all of this was going on, Bob's strength and stamina bars were up on the screen. Both bars were empty and the drivers name was Damon. Non of my drivers names start with Damon if that means anything. Needless to say, I rebooted and our league race ended up having to be pushed back in hopes that the next update would fix some of the issues.

Edit: Our room room always had at least 14 players trying to race.

 

[chukabilly]

I'm confused. Either PD is already aware of the issue (like some important members seemed to imply) and then we don't really have to gather anything, or they aren't. In which case all we really should have to do - run a bunch of races with >13 players repeatedly and save the replay.

Famine's post confirms that PD is aware of the issue, and that "important members" themselves are trying to gather data that points to a solution, rather than keep reiterating a known problem.

Our purpose right now is continuing to gather information that they can use; to either propose, or support, a fix to the problem.

PD is not revealing any of their thoughts on what that fix might be, so it is up to us to provide the Admins with information that is relevant to that end, namely: fixing the glitch!

Pd needs to be ready to admit that only ten people can race in a room, or come up with a solution that works for sixteen people. Since they desire a fix that enables the full field of 16, they will naturally look to achieve that first. But that takes time. Time they don't have, which is why it's even more important now that we continue to post our results from all of our investigations, regardless of the series. {I'm partial to WRS!!!}

Did someone mention TIME?
I am investigating the clocks used by both PC's and gaming consoles. It has, of course, to do with the shape of the silicon quartz used to produce the oscillations counted every second. These oscillations are counted by two separate tools: the software clock and the hardware clock.

It is in combining these two clocks on every console, and then adding them to every other clock from every other console {with every console actually being two clocks}, that the problem keeps arising, WITH MORE THAN TEN PEOPLE!! What is it that the eleventh person does to upset the system?

I believe that number 11 distorts the clock, which affects the physics, which affect lap time. As I have mentioned before, the PD priority is fixed on the visuals, not the physics. Therefore, the lap will look good, it just won't be fast. [Kaz is a photographer at heart, after all!]

As to exactly what the eleventh person is doing to the clock, I can't explain right now because neither Sony nor PD are making the exact shape of their quartz available to the general public. I will continue to work on that!

My principle note to all who read this is that there seems to be a coincidence between the number 10 and 11, and the clock we are all using is based on the number 10. The Real Time Clock, {RTC}, should be aiming for 32.768 kHz per second, which is pretty precise, so how exactly can we "synchronize our watches"? And what can we do to make sure that a race we want to run is accurately measured by one clock, not twenty in a ten person room, or thirty-two in a 16 person room?

Was the fatal mistake made in measuring Time on a TENTH scale and then extrapolating to a SIXTEENTH scale? That won't work. Computers are binary. The number 11 is a prime number, and the first the computations are forced to reconcile above the number 10.

The 32.768 kHz number equates to 0b1000000000000000, with a hexidecimal of 0X8000. Just because the hexidecimal is both a divisor and multiple of 16 does not mean that it will translate into a time recording system based on the number 10.

I truly believe that this is the problem. There is not enough computing power in the PS3 to instantaneously convert binary code to "real-time" physics every second of every race, especially if you want to demand the program be available to immediately satisfy every request made by any and all outside influences, such as someone jumping in to watch the race. Let alone handling people in Australia racing with people in Europe racing with people in the Americas (both North and South). And if there is enough computing power in the PS3, then PD isn't using it right!

There needs to be a reconciliation between RTC's and game time clocks. Maybe we need to set our clocks via internet, and then go directly to the room we want to race in, providing everybody does the same thing at the same time.

And I still hold to the opinion that security issues are paramount, so the solution must be compatible with whatever the new security protocol is.

And I continue to believe that a lot of this can be fixed if PD will loosen some reigns and let us choose Hosts who can isolate a room, so at least we know that whoever races there are on a level playing field!

Racing On!


edit: Owens! 👍
edit: added the spaces for you chuk.

 

[Famine]

The only way around it is to limit the lobby to 10 people. We are several people not bothering to post here because _all_ the sufficient information is already here.

The Gran Tursimo box say its capable of a 16 players online experience, it's not. It's not the owner of this forum or the users of this forum to figure out why. PD are the only one who could fix this.
We could sit down and speculate for months or even years so if you don't hesitate to post. Well it's about time.

Funnily, a prolonged test last night with several GTP_ drivers and staff showed this statement to be incorrect. The test did reveal specific circumstances that caused the bug (for whatever reason) and specific ways of avoiding it even with a 14 man lobby. We also had a 16 man lobby with players from the USA, Canada, Belgium, the Netherlands, the UK, Finland and New Zealand all at once. Looks like the game is capable of a 16 player online experience like it says on the box.

So we could sit down and speculate for month or even years, or we could just post up rubbish without checking our facts. What we've chosen to do is work out the issue and come up with a workaround for members and the larger GT community.


As Jordan said, a blog post is not necessary to get issues fixed - few bugs have been blogged before they were fixed. He also said that we would make a blog post about it when we have enough information to help everyone. It is time when the information is correct and no sooner - this is why GTP has sufficient credibility to be invited to private Sony events despite being a fansite.

 

[jon-]

Funnily, a prolonged test last night with several GTP_ drivers and staff showed this statement to be incorrect. The test did reveal specific circumstances that caused the bug (for whatever reason) and specific ways of avoiding it even with a 14 man lobby.

So are you going to share how to avoid it with us or just continue to preach while all our online racing is effectively ruined?

 

[Famine]

So are you going to share how to avoid it with us or just continue to preach while all our online racing is effectively ruined?

As Jordan said, a blog post is not necessary to get issues fixed - few bugs have been blogged before they were fixed. He also said that we would make a blog post about it when we have enough information to help everyone. It is time when the information is correct and no sooner - this is why GTP has sufficient credibility to be invited to private Sony events despite being a fansite.

Time spent having to restate the same things to the same people is time not spent getting the information we need.

 

[jon-]

Try and see it from our point of view. It's Sunday today, and a LOT of online racing leagues race Sundays so in around 10 hours I'll be sitting down in a 16 player room hoping things run smoothly.

You said you've found a way of running busy rooms without the bug. This is great and we're all very thankful for your work.

Even if you're not 100% sure of it being a fix, PLEASE share what you think might fix it with us, that was if we're affected by the bug tonight, we can at least try your proposed solution and feed back the results to you.

Many people trying your fix is far better than one room trying your fix for validation. Also most of us are happy to try ANYTHING to get a good race.

I'm not asking you to blog about it, or ruin your well earnt credibility with Sony, just simply say:

"Try this, it might help...."

For the community that actually pays your bills. We won't care if it doesn't help, but we'll damn well love you if it does.

 

[Famine]

Try and see it from our point of view. It's Sunday today, and a LOT of online racing leagues race Sundays so in around 10 hours I'll be sitting down in a 16 player room hoping things run smoothly.

We do. The GTP_WRS is a significant part of this site. There is no us vs. you - we're all part of the same community.

You said you've found a way of running busy rooms without the bug. This is great and we're all very thankful for your work.

We've found correlations. They need to be checked. The people doing the checking are US West Coast - currently just after 4am.

Even if you're not 100% sure of it being a fix, PLEASE share what you think might fix it with us, that was if we're affected by the bug tonight, we can at least try your proposed solution and feed back the results to you.

Many people trying your fix is far better than one room trying your fix for validation. Also most of us are happy to try ANYTHING to get a good race.

We're getting there. I'm writing the article now (or I would be).

I'm not asking you to blog about it, or ruin your well earnt credibility with Sony, just simply say:

"Try this, it might help...."

For the community that actually pays your bills. We won't care if it doesn't help, but we'll damn well love you if it does.

What community pays our bills? GTP staff are volunteers on our own time...

This needs to be right. We need to put up one, unequivocal post that says "If this problem occurs, this/this/this and/or this will work.". We don't need to give out misinformation and if we get it wrong that's exactly what will happen.

Ultimately, until (unless?) the bug is fixed, no temporary workaround we can provide will be satisfactory.

 

[jon-]

We do. The GTP_WRS is a significant part of this site. There is no us vs. you - we're all part of the same community.

No arguments here.

We've found correlations. They need to be checked. The people doing the checking are US West Coast - currently just after 4am.

1 room checking the correlations is good, 10 rooms checking things for you tonight is better.

We're getting there. I'm writing the article now (or I would be).

In your situation I wouldn't want to post an article until I'd had confirmation from MANY sources. Post the work around to this thread first and allow us to validate things, at least anecdotally for you.

What community pays our bills? GTP staff are volunteers on our own time...

I understand the moderators do a difficult job for free, and fair play to you, but at the end of the day a site as popular as GTP will be cash flow positive for the owner thanks to the advertising, which wouldn't be possible without the community.

This needs to be right. We need to put up one, unequivocal post that says "If this problem occurs, this/this/this and/or this will work.". We don't need to give out misinformation and if we get it wrong that's exactly what will happen.

Correct, for the blog post you need to be 110% sure otherwise GTP might look a little stupid. Given the huge amount of variables in online racing I still find it hard to believe you can get to 110% with just 1 team of testers. Using the dedicated racers in this thread however you can at least get closer to 100%.

Ultimately, until (unless?) the bug is fixed, no temporary workaround we can provide will be satisfactory.

Incorrect (IMHO) - I would be very happy with a workaround which mitigates the problem, however slight, for our racing. Only PD can provide a bug fix.

 

[Speedy6543]

I'd suggest all you guys here, instead of hoping the full grid races go well, to do some tests of your own and post the results in here.
Laps run with 1st lap specs, being all packed up together.

• Using wheels + mics
• Using wheels only (no mics)
• Everybody switching over to controller (DS3/Sixaxis) + mics
• controllers only (no mics)

Comparing differences (laptimes/tyre grip) afterwards.

EDIT:
possibly all these testing can also be done with people dropping out lap per lap 'till you get to around 10 people still 'racing' and see how that evolves.

 

[Famine]

At least we're mainly on the same page now - I can't speak for Jordan's finances and I'd suggest you don't either - but we still have the same problem. Anything I say, as a GTP Admin, is "official" and if it's wrong I and we look stupid.

We'll get something sorted well before Sunday night races. It's that or watch a Top Gear repeat and I don't think anyone wants to do that in case it's the India Special again.

 

[jon-]

I'd suggest all you guys here, instead of hoping the full grid races go well, to do some tests of your own and post the results in here.
Laps run with 1st lap specs, being all packed up together.

• Using wheels + mics
• Using wheels only (no mics)
• Everybody switching over to controller (DS3/Sixaxis) + mics
• controllers only (no mics)

Comparing differences (laptimes/tyre grip) afterwards.

We have, if fact a room of us did Friday and it echoed what someone posted above in a neat spreadsheet view.

With all things being equal you gain time by turning your mic off, and then gain a little more time switching to DS3.

 

[jon-]

We'll get something sorted well before Sunday night races. It's that or watch a Top Gear repeat and I don't think anyone wants to do that in case it's the India Special again.

- there isn't even a repeat on the bbc tonight.

Thank you for committing to have at least something up to test before the end of the day. Will make sure our room does some detailed testing tonight if the bug comes up (we're running 16 people, stock premium SLS from the garage with only paint modifications to wheels and body on sports mediums and hards - we've all practised enough to know our times)

 

[esh]

^While the issue isn't new, it wasn't too bad until recently. So I doubt it's just the priority of visual over physics. Something clearly made it worse, so it could/should be feasible to undo/fix.

I vote for the adjustment to the sound system.

 

[esh]

The only way around it is to limit the lobby to 10 people. We are several people not bothering to post here because _all_ the sufficient information is already here.

The Gran Tursimo box say its capable of a 16 players online experience, it's not. It's not the owner of this forum or the users of this forum to figure out why. PD are the only one who could fix this.
We could sit down and speculate for months or even years so if you don't hesitate to post. Well it's about time.

Fortunate for you there's enough people around here that aren't interested in just having some idea that there is an issue and actually are trying to understand every nuance of the bug as well as ways that could provide a workaround sufficient enough to make a news post once the issue is as close to fully understood as possible.

Why is this such a difficult concept to understand? Why do so many people just want an answer or someone to tell them that everything is going to be okay instead of doing a little research themselves? I'm not necessarily singling you out but I have not seen a single post from a moderator or admin @ GTPlanet that states anything other than "we've got your back" and all people still want to do is make trite assumptions about their and this site's motivation or lack thereof.

 

[ErikAndre71]

Fortunate for you there's enough people around here that aren't interested in just having some idea that there is an issue and actually are trying to understand every nuance of the bug as well as ways that could provide a workaround sufficient to make a news post once the issues is as close to fully understood as possible.

Why is this such a difficult concept to understand? Why do so many people just want an answer or someone to tell that everything is going to be okay instead of doing a little research themselves? I'm not necessarily singling you out but I have not seen a single post from a moderator or admin @ GTPlanet that states anything other than "we've got your back" and all people still want to do is make trite assumptions about their and this site's motivation or lack thereof.

It was not my attention to assume they have lack of anything. But even if there are some game developers, network experts or bug experts here it will fall down to speculations. Yes maybe one of those many will be lucky and find the answer to this.
But only people who work on this game sits on the game code itself, none of those self announced experts on this forum does. I try to solve this myself, I first tried everything that made sense to me and then I tried things that pupped into my head that didn't make sense at all. But I gave it a try because we came to a place we are close to give up.

If you mean I assumed this sites lack of motivation, I think it's fair to say you assume I'm not trying to research this.

We did some tests already, but when I was about to post facts about it. It was already done. I really don't see why a double post will solve anything.

I agree that someone here seems to sit and wait for a answer or a PD update to fix this. And now I do the same thing. I tried everything I could think of and even if I love this game and this site, I'm not going to waste more of my time trying to solve a problem that I can't.

 

[esh]

It was not my attention to assume they have lack of anything. But even if there are some game developers, network experts or bug experts here it will fall down to speculations. Yes maybe one of those many will be lucky and find the answer to this.
But only people who work on this game sits on the game code itself, none of those self announced experts on this forum does. I try to solve this myself, I first tried everything that made sense to me and then I tried things that pupped into my head that didn't make sense at all. But I gave it a try because we came to a place we are close to give up.

If you mean I assumed this sites lack of motivation, I think it's fair to say you assume I'm not trying to research this.

We did some tests already, but when I was about to post facts about it. It was already done. I really don't see why a double post will solve anything.

I agree that someone here seems to sit and wait for a answer or a PD update to fix this. And now I do the same thing. I tried everything I could think of and even if I love this game and this site, I'm not going to waste more of my time trying to solve a problem that I can't.

Doing research and providing input, no matter how seemingly insignificant, does not equal trying to find a solution. It's to promote understanding. I would like to understand ways in which I can reduce my frustration with this game and issue so I try to glean as much info from this thread as I can, as well as my own experience, and remain positive. That state of mind may change 3 months from now if the issue still persists but that would be putting the proverbial cart before the horse. A common practice here which is evident in this thread.

And again, it wasn't my intention to single you out. There had been a rash of posts up to this point, again, that required Famine to apply his subtle diplomacy skills, which is always a treat to witness. I just made an observation.

 

[ErikAndre71]

Then I must say I appreciate you guys to keep going. I used this game to relax from my daily research and problem solving. Not going to expand those hours.

Also I liked this [post=6506074]post[/post]. Good read!

 

[BrandonW77]

We did some tests already, but when I was about to post facts about it. It was already done. I really don't see why a double post will solve anything.

We need all the data points we can get. If you have some recorded results you should post them here even if they are the same as other results. Anything we can do to confirm when the bug does or does not happen is valuable information. 👍

 

[ErikAndre71]

We need all the data points we can get. If you have some recorded results you should post them here even if they are the same as other results. Anything we can do to confirm when the bug does or does not happen is valuable information. 👍

I search my computer, but I think I forgot to save the notepad document I used for the data.

We was about to start our ICGTC season 3 at Fuji when it all started. Qualify went fine. No issues what so ever.
We tried both fixed and none fixed ownership, I'm on a fiber optic 60/60mbit connection so a fixed ownership with me as the owner should work. We red flagged this race so I have no data from it. No replay's of a aborted race.
Then we went into a none fixed ownership room and lag continued. We restarted the race a few times and in the end we decided we cancel the race. We did however finish this race and I got myself a replay. This race lasted for 38 laps around fuji and I'm sorry to say I wont go in and do the job all over again.
We did not change between wheel or DS3, but we did the first 5 laps with all mics turned off. And we do have a few who only use DS3.
The data I got from the replay was pretty close to MGRs [post=6491337]post[/post]
We also had a wheel user who didn't seem to be affected at all. At least not on the replay and the time he managed to do. (same as in free run)

During the race I heard people say pit crew wasn't there. The tyres got changed, but no pit crew to do the job. People who qualified with a laptime of 1:35 did lap times of 1:40-1:45 with good tyres.

I noticed that my gear sometimes wouldn't change, had to change it 2-3 times before my gear box actually changed gear. Steering was none responsive at times

Also the games frame rate was way lower than it should be.

After 10-12 laps some guys gave up and when we got below 12 people the race got closer to normal.

 

[ADEuk]

Since the latest update I notice much more lag when switching from shuffle to PP racing. I wonder if it has anything to do with painted rims on standard cars?

 

[t_w_a_h]

I have been experiencing this more and more and have noted my own experiences below; anecdotal as it may be, perhaps it will help.

User Context

Internet Connection: Wired - 10 to 13 mb download - 0.8 to 1.3 mb upload
PS3: Slim - 320GB - CECH2053B
Controller: DS3 and wheel
GT5 Online Use: Shuffle racing, recommended car garage/same car racing, occasional organised league races with relevant regulations

Online Experience

My usual GT5 online racing is done with the same people; their internet connection vary wildly from great fibre connections to fairly low speed rural connections (lowest noted 5mb down/0.5mb up), there are regulars from the UK, Germany, Switzerland, Australia and New Zealand, about 50% use mics.

When we shuffle race during the week with the room usually set at a max of 14 we don't often experience any problems. Traditional lag appears now and then with the usual signs, cars jumping around the track and mics becoming stuttering. This lag shows in replays where the cars can be seen stuttering. This is almost always caused by a[nother] distant overseas / low speed internet player joining the room. Frequency has increased since update 2.0.

We have also experienced an increase in invisible car syndrome where several players in the lobby have a lower number of total racers in their race than the rest of the room. Player A sees 10 in the race but Player B (who is not shown on the race list but is shown on the lobby list for Player A) has 12. This leads to the Player B being rammed by Player A though no effect is shown on Player A's screen.

We can however, during the week, more often than not, run shuffle racing with 14 players in a lobby with the regulars, including the continental Europeans and Antipodeans with no problems.

On a weekend we do a longer league race using either a car from the recommended garage or an untuned [premium] car from our own garages. Since 2.0 these races have been blighted by slow cars and lagging frame rates. This has become know to me as the "driving through icy treacle" effect. The ice and treacle are somewhat of a paradox I know but that's honestly the best way I can describe it. The track itself seems to stutterer by slowly whilst the tyres seem to lose all grip. Lap times increase exponentially and even straight line speed is lowered.

This has been noted in private and public lobbies made up of the same people and in rooms of both types hosted by different hosts.

This causes almost all races to be restarted at least once and usually at least 2/3 to have to drop out. Once the room size drops to 11/12 the problems seem to disappear or at least become largely unnoticed. Both wheel users and DS3 users seem to have trouble but it is more apparent and appears to be more frequent amongst wheel users. Changing from wheel to DS3 during the race doesn't appear to directly cure the issue but may have an effect (placebo perhaps?). Weirdly the low speed internet connection users don't seem to have it happen more than those with high end connections though they are more likely to appear as traditional laggers on other users screens. During a recent race [all premium] where this was being noticed I did some lap timings on my stop watch versus the game clock was never more than .4 off (reaction times I expect). Unfortunately I was disconnected before getting the chance to save a replay so was unable to time the same laps.

Possible Inferences Regarding Slow Laps

- The internet connection of the user experiencing the slow down may not be an issue.
- The poor internet connection of others in the room may be an issue.
- The host internet connection may not be the primary issue. We have tried all users as hosts and traditional lag appears to increase when the host has low down/up speeds but the slow lap glitch doesn't appear to increase in frequency.
- Room size/cars on track may (I would personally say this is a definite) be a factor.
- Control method may increase/decrease frequency of the problem.
- General PSN traffic may be part of the problem (weekday vs weekend usage).
- The additional detail of premium cars may be a factor (shuffle vs recommend car garage).

Not really much help I know!

What I will say though is that it has become such an annoyance that last night after a few restarts I just gave up and went and watched some telly. After what was nearly a year of reasonable performance it's starting to really get annoying. And annoying enough for me to start searching for alternative online racing experiences to play in my limited gaming time.

 

[MGR]

I never really noticed this bug until I started doing some 'organised' racing on tracks without boost turned on. My first WTF moment happened in October last year when we tried to run a 14 car race on Deep Forest using tuned Fiat 500's. Prior to that we'd just been mucking around doing one-make races from the arcade (recommended) list with boost turned on.

Wind the clock back to June last year....

I had recorded and kept a bunch of replays from an evening GT5 session. These rooms were run with boost on HIGH so this bug wouldn't have been very easy to spot given the monumental speed difference between the first and last car in a HIGH boost race.

Here is a sample of the lap time differences recorded from some of the replays on the 16th June 2011:

- The race with 9 competitors seems OK.
- Bump it up to 12 and now the bug shows itself.
- Raise it to 14 or 15 competitors and the game is imploding on itself.

Unfortunately I don't have any replays prior to June 2011 so I can't check back any further. But the prospect of a fix - short or long term - is not looking very good right now.

 

[dudeondacouch]

Ran a few races in a public lobby last night.

Shuffle races with 14 people seemed to be working pretty well (better than usual in fact), then we switched to a one-make with a premium car. Even with no additional people, it was a complete tragedy; room became unusable. Back to shuffle and everything was ok again.


It could have had something to do with a track change, or another variable, but I think this would be a worthy test to run.

Fill a room and run different race types with standard cars, then premium cars, with exactly the same people, and check replays and feedback.

 

[cicua]

@MGR - good work ... what I guess we suspected/feared. GT5 online timing is totally unreliable and always has been.

I believe that if/when they fix this, online/offline physics will suddenly become the same as well. These things seem connected. I know others said this before, but now it seems even more plausible.

 

[Furinkazen]

I'm running shuffle races as they appear to be only races that are not badly affected.