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

[Jake]

Chuk!

 

[cicua]

Thinking about Time affecting the physics engine made me curious as to my own date and time setting, so I looked at it. It read 11:53. I noticed my cable box said 11:49, and it tweaked me! So I updated via internet, and Presto! Now the PS3 had the same time as my cable box. They ALWAYS should, because we have a bundled package of cable/internet/phone service.

All computers have a hardware clock, a chip that keeps ticking at very regular intervals. The computer OS bases it's idea of current time on this. But hardware clocks are not accurate. Due to heat and other stuff they may not tick exactly every millisecond, but every 1.0000000001 millisecond, say. And so they tend to drift over time (this is where your 4 seconds come from), so they need to be synced up to some official time every now and then.

The best protocol for time syncing is NTP. I believe this is what the PS3 uses when syncing "over internet" (again it is beyond me why it does not just do this automatically like every modern PC).

Interrupt requests are the "pings" from the hardware clock, or PS3. It should be every 54.936 milliseconds, or 18.206 times per second.

I hope your info is outdated, 18.2 ticks/second was used in DOS PC's of the 80's and 90's, but 18.2 ticks/second does not provide very much precision. PC hardware has got much better timer chips since then and I sure hope the PS3 has too.

This is new to me. The clock could actually really be the source of the bug? I mean, it's just a clock Will give it a try anyways.

All the GT5's in a room need to synchronize their clocks. How else would they know how to start the race at the same time? This does not mean they sync their system clocks, but GT5 must have a relative clock running (the race clock, starting when the track has loaded), and this needs to be synced between players.

This sync'ing should not be based on system clocks. But all software history shows that the gap between "should" and "is" can be infinitely large.

Time syncing over network is far more hairy than one would think. You need to take latency, jitter and packet loss into account if you want any kind of accuracy. This is what the NTP time-syncing protocol does so fantastically well, it can achieve millisecond precision even over not very good links.

But this being GT's first-gen online code, and GT not being a time-syncing application primarily, I could easily imagine that the developer tasked with this was not the biggest time-syncing-over-network wizard, and that (s)he might have introduced a dependency on the system clock, for whatever misguided reasons.

At least, that is what I want to find out.

Sorry for wall of text. I am a nerd and this triggers my curiosity.

 

[Ealirendur]

Chuk, I personally think it unlikely it will be a factor, but the time sync mechanism is likely to be NTP or the simple variant.

EDIT: Partially tree'd by Cicua

I can check to see that it is later, if you like (by directly checking whether the PS3 is in fact doing NTP requests over the network when updating the time).

I have a great deal of network data from a few races (including the 3D3 tests), but useful results from analyzing it are difficult to realize. Interesting, though, and I may post the findings, but they're purely technical at this point (trying to find useful graphs given the time frames of the races/tests is tricky and time consuming).

Short summary: most of the traffic is very small UDP packets, of which the frequency varies by room quality, and is partly proximity dependent. Mic traffic can easily be approximately the same amount when in use, and is probably encrypted.

The mesh is also controlled by UDP, individual racer's connections are seemingly evaluated by service.playstation.net, but would seem to be updated/rechecked between the peers every minute with small updates/keepalives/whatever every 10 seconds.

Bandwidth would be rarely a problem (mostly upload), but I can definitely see where latency/round trip times could be.

I still would very much like to see a test with the same 14-16 people with both very low and extremely high room quality settings, with mics off; I will endeavour to procure them at some point if no one else does so.

 

[chukabilly]

clock bug squashed

I am outdated, but still looking!

edit: I am reading and digesting your posts, cicua and REN!

 

[Astro_BS-AS]

^ Don't think framerate and the glitch are too related. Had many instances of slow framerate with no noticeable change in speeds.

Hi.

The problem here semms to be different to the typical framerate problems.
Here the game "is missing", not those frames, but the actual "processing of the car" in the track.

I read the complete post and I think its amazingly well explained. At least now I realize I was not that bad when racing haha. It happens to me always using a DFGT.

 

[jon-]

Here's something (sort of) interesting - last race we did showed the usual pattern.

Wheel users hit
Pad users not hit

With 1 exception. The only person using a Fanatec wheel (Forza Motorsport CSR Elite Wheel with CSR Elite Pedals) wasn't hit.

Now I'd have assumed each wheel would process the input data into a common code for the ps3, but perhaps the Fanatec is doing more of the heavy lifting and leaving the PS3 less to do? He used to run a G27 and was always one of the worst hit with the bug.

Probably nothing, thought I'd mention it anyway.

 

[cicua]

Now I'd have assumed each wheel would process the input data into a common code for the ps3, but perhaps the Fanatec is doing more of the heavy lifting and leaving the PS3 less to do?

Certainly a possibility. All hardware is different, just because they generate the same commands does not mean the act the same. Some might behave good, others bad, like sending commands too fast, too slow, in wrong order, etc.

Does the Fanatec use a dongle so it pretends to be something else? If so, maybe the dongle behaves better than a "native" wheel.

Wheels in general might generate many more interrupts (commands) for the PS3 to process than a DS3 does.

Another thing I just thought of: We got "official" G27 support in spec2, that means they have changed something in the wheel/input code. A change of behaviour in the input layer, or just a tiny little bug sneaking in here, might be the cause for the whole think (if so nothing we can do anything about).

 

[RUFin]

Another thing I just thought of: We got "official" G27 support in spec2, that means they have changed something in the wheel/input code. A change of behaviour in the input layer, or just a tiny little bug sneaking in here, might be the cause for the whole think (if so nothing we can do anything about).

This.They dont like Logitech and dont want G-wheels.You can turn your car with it but never again be real fast with it.Goes how it feels too.I have had these slow downs and lost of grips since the end of Prologue(maybe from day 7?).Im one off those (probably) very fews who drow Prologue all aids off(ABS too) incar and clutch,already after first couple months.My consol has been monitored agese now.Every time I stop playing GT it saves,even in offline.Sometimes when I just use its internet.Im on their 'list'.Been there as a test bunny and had these mentioned problems a LONG time now.Why they do it,its easy.Want to cash in every frontier.Money talks,simple as that.And again,what happened to Beginner,Intermediate,Expert options?In there,all together,just dont know who has what.You cant upset young players to let them feel 'I cant drive'.-Driving,NOT racing.Harmonised online,without you even noticed it,thats their after./100% sure guess.

 

[Duncs]

"official" G27 support in spec2

Im pretty sure this just meant we could configure all the buttons and made no other changes, I would be surprised if this had any effect on actual gameplay as most other wheels already had this function. Who can tell though, im no expert for sure !

 

[RamoSapiens]

Wait a minute...

When I changed from Fanatec to G27 i was suddenly competitive against
Guys like Nissman, and actually beated Tony once I had also incredible
Good tire wear....! With same car, same setup i had much better tires in tests...?


Now when Im back to Fanatec again, Im 1 sec slower, and tire wear is "normal"
Compared to others.

 

[EDK]

If PD didn't like Logitech and didn't want "G-Wheels", why would they give an update to support them with mapping?

I will double check with those we had present, but I think we had just about every type of wheel device imaginable present during our test session last weekend.

I use a T500RS, there were definitely G27 Wheels, at least one DFGT, at least one G25, and probably even a DFP. I'm not certain as to whether there was a Fanatec wheel in the room, that's the only one I'll have to double check on.

 

[jon-]

EDK, likewise I'm on the T500 RS and we had a mix of G27, G25 and DFGT. Only the Fanantec guy managed any sort of pace, the rest of us were 8 seconds off the pace for the first few laps, then most of us stabilised at around 3-4 seconds off the pace.

 

[EDK]

EDK, likewise I'm on the T500 RS and we had a mix of G27, G25 and DFGT. Only the Fanantec guy managed any sort of pace, the rest of us were 8 seconds off the pace for the first few laps, then most of us stabilised at around 3-4 seconds off the pace.

We found that anyone who was out front in clean air had pace.

Now that I'm recalling who all we had in there, I do know that we had at least one Fanatec Wheel present, and he was not just running away with it.

That said, I'm going to work to compile a complete and accurate list of the devices being used during our test.

 

[dsgerbc]

With 1 exception. The only person using a Fanatec wheel (Forza Motorsport CSR Elite Wheel with CSR Elite Pedals) wasn't hit.

Now I'd have assumed each wheel would process the input data into a common code for the ps3, but perhaps the Fanatec is doing more of the heavy lifting and leaving the PS3 less to do? He used to run a G27 and was always one of the worst hit with the bug.

Probably nothing, thought I'd mention it anyway.

Yay! Now I can join the 'there is no bug crowd'. Seriously though, CSR-E pretends to be g25 to PS3, so what you're telling is likely a fluke. Or a glitch in g27 implementation. I fairly certain that in most online races there's a least one g27 present.

A can report back though, as a former g27 user who used to get hit by this with high probability in rooms with more than 12-13 drivers. I'll jump into some random crowded rooms tonight and see if my new wheel is helping me.

 

[Driver321GO]

I also had some issue with the time and lag.
Sometimes if a race starts, the other drives are 3 -6 sec above from me until my start signal appears, so maybe this is related to a lag issue or time sync problem...?

 

[dsgerbc]

^No. Clean your network cache before races.

 

[EDK]

Here were the users and wheel devices in our test. 3 G27's, 5 DFGT's, 2 Fanatec Wheels, 2 G25's, 1 T500RS, and 1 TBD.

w-g-e – Fanatec GT3 RS Clubsport
Ealirendur – G27
MilleRSVR – Fanatec GT2
Hasslemoff - DFGT
EDK – T500RS
Famine – G25
Segovia - DFGT
MadMax86 – G27
Hydro - DFGT
PASM – G25
Speedy6543 - DFGT
Sjaak – G27
Gravitron - DFGT

Every driver was affected by the bug at some point during the test process, so I don't believe a correlation can be formed between brand/model of wheel device and the bug.

 

[jon-]

That doesn't surprise me, but thanks for checking.

You need to get a DS3 user in a test. They fly!

 

[BrosifDuder]

This.They dont like Logitech and dont want G-wheels.You can turn your car with it but never again be real fast with it.Goes how it feels too.I have had these slow downs and lost of grips since the end of Prologue(maybe from day 7?).Im one off those (probably) very fews who drow Prologue all aids off(ABS too) incar and clutch,already after first couple months.My consol has been monitored agese now.Every time I stop playing GT it saves,even in offline.Sometimes when I just use its internet.Im on their 'list'.Been there as a test bunny and had these mentioned problems a LONG time now.Why they do it,its easy.Want to cash in every frontier.Money talks,simple as that.And again,what happened to Beginner,Intermediate,Expert options?In there,all together,just dont know who has what.You cant upset young players to let them feel 'I cant drive'.-Driving,NOT racing.Harmonised online,without you even noticed it,thats their after./100% sure guess.

Please use the spacebar after periods and commas. It is very hard for me to read your post. Thank you.

 

[OwensRacing]

clock bug squashed

I am outdated, but still looking!

edit: I am reading and digesting your posts, cicua and REN!

I lived through that!!! I had serious separation anxiety from COD then. LoL


That issue, maybe still plague us on a smaller scale. Not a crippling scale like that was.

Then again I truly don't know and am guessing.

 

[EDK]

That doesn't surprise me, but thanks for checking.

You need to get a DS3 user in a test. They fly!

Right, we know. We did that as well. Basically, as soon as wheels and/or mic's were removed, things got better. As soon as we went to 12 drivers, things got better.

We performed a broad and thorough test, in an environment that was arguably as static as possible in terms of performance (Oval that could be taken flat out the entire time).

 

[DA__dogg]

This bug is such a pain for online lobbies and leagues that have more than 12 racers. I'm running a league right now that had 15 people in the race at Monza last night. All but maybe two or three people experienced drastically different lap times in the race compared to the free run in qualifying. Looks like I have no choice but to limit the races now to 12 people. I really hope PD get's on this right away if they aren't already.

 

[EDK]

This bug is such a pain for online lobbies and leagues that have more than 12 racers. I'm running a league right now that had 15 people in the race at Monza last night. All but maybe two or three people experienced drastically different lap times in the race compared to the free run in qualifying. Looks like I have no choice but to limit the races now to 12 people. I really hope PD get's on this right away if they aren't already.

We've dealt with it in our Official WRS Online events. In some ways, it gives more flexibility, as it will allow you to have up to 24 participants.

See the bracket breakdown in this post and all of the related info in the thread.

You just have to think of things differently. But trying to run with 15 people in the room is like trying to get by a brick wall by running through it. Try jumping over it or running around it.

 

[jon-]

^^ We're going to try and run a 15 player race in free run this weekend. I'm not entirely sure it's a good idea but that's what they want to do

Lots of trust involved.

 

[DA__dogg]

Thanks for the link but that league is WAY more organized than ours is. We are just a bunch of people that met online over at the Sony Forums. Last year we had maybe 5-8 people per race. Now we are in season 2 and suddenly 16 people want to race with us. That is great I'm not complaining about that but the lag and time difference is absolutely ridiculous it's not fair. If limiting the races to 12 does not work then I don't know what to do. Clearly your league has the resources of GTPlanet. Kudos on that.

 

[EDK]

Thanks for the link but that league is WAY more organized than ours is. We are just a bunch of people that met online over at the Sony Forums. Last year we had maybe 5-8 people per race. Now we are in season 2 and suddenly 16 people want to race with us. That is great I'm not complaining about that but the lag and time difference is absolutely ridiculous it's not fair. If limiting the races to 12 does not work then I don't know what to do. Clearly your league has the resources of GTPlanet. Kudos on that.

What we did is not as complicated as you might think.

16 people are racing. Split them into 2 rooms of 8 each, with relative ability spread evenly across the 2 rooms. Then run 2 separate races.

The top 4 finishers from each race meet in a final race to determine overall winners. The lowest 4 finishers from each meet in a "consolation" race.

Yes, we have great organizers. But we all started out as participants, just like you, and most of our resources are home grown.

 

[DA__dogg]

Thanks for helping me. I am not ashamed to say that I am suddenly feeling overwhelmed at the possibility of people quitting my league because of this. I apologized to everyone last night and quite a few people left soon after the race without saying anything. I really felt bad. If you don't mind, I'd like to use the template of your league and implement it into the Exotic Touring Series, the league I run. I'll do my best to keep the league together. If nothing else I'll just make it a 12 person league with it being on a first come first serve basis. Banning wheels and mics is not a possibility imo. I use a wheel and do not want to go back to a controller lol. Thanks again.

 

[BrandonW77]

Thanks for helping me. I am not ashamed to say that I am suddenly feeling overwhelmed at the possibility of people quitting my league because of this. I apologized to everyone last night and quite a few people left soon after the race without saying anything. I really felt bad. If you don't mind, I'd like to use the template of your league and implement it into the Exotic Touring Series, the league I run. I'll do my best to keep the league together. If nothing else I'll just make it a 12 person league with it being on a first come first serve basis. Banning wheels and mics is not a possibility imo. I use a wheel and do not want to go back to a controller lol. Thanks again.

Banning wheels isn't an option really. But it would probably be a good idea to disable chat for now. I don't like doing it either but there's lots of evidence in this thread showing that having mics off and limiting to 12 people are the best things we can do to minimize the effects of the bug. I recently disabled mics (in the lounge settings) for my league and we seem to be having fewer problems.

 

[EDK]

Thanks for helping me. I am not ashamed to say that I am suddenly feeling overwhelmed at the possibility of people quitting my league because of this. I apologized to everyone last night and quite a few people left soon after the race without saying anything. I really felt bad. If you don't mind, I'd like to use the template of your league and implement it into the Exotic Touring Series, the league I run. I'll do my best to keep the league together. If nothing else I'll just make it a 12 person league with it being on a first come first serve basis. Banning wheels and mics is not a possibility imo. I use a wheel and do not want to go back to a controller lol. Thanks again.

re: The Teplate.

If you want to use our organization method, have at it.

Please don't use the images.

And please don't leave verbiage in that would suggest your series is an officially sanctioned GTP event.

 

[Speedy6543]

Banning wheels isn't an option really. But it would probably be a good idea to disable chat for now. I don't like doing it either but there's lots of evidence in this thread showing that having mics off and limiting to 12 people are the best things we can do to minimize the effects of the bug. I recently disabled mics (in the lounge settings) for my league and we seem to be having fewer problems.

The bold text + Race quality standard and visible damage [off].