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

  • Thread starter MGR
  • 1,131 comments
  • 96,766 views
I have really noticed this after 2.2 I was on spa just practicing in some lmps and I was running 10 second slower then what I usually do. There was about 14 people in the room. I was not making mistakes and I was running very clean and consistent, but I could not get better the a 1.56-1.57. I use a the force feedback wheel but I didn't think that would be a issue, but now I guess it could be.
 
Great post Sail, pretty much the nail on the head.

As mentioned I'm more inclined to think it's related to the playstation load rather than network. We only race with UK racers, we're all on pretty decent connections and GT moves a comparatively small amount of data.

What makes me think it's graphics / cpu related is it's far more apparent on what I'd call "high draw" circuits, stuff like SSR5 with lots of detail, or Daytona Road with lots of wide open spaces. Also it seems worse when running multiple premium cars.

Sail, if you've any replays saved where it was really bad get a stopwatch and time your laps against what the GT5 replay thinks you're running, then time the same race from someone who wasn't affected.

During testing we found a fairly big difference between stopwatch and GT5 laptimes for those who were affected, unsurprisingly around the same times as we were losing. It's almost like the internal timers get messed up as the cells get loaded.
 
I experienced the same problems aswell. On the free run session i was able to race bumper to bumper with the host of the room i was in and we had some really nice overtaking and blocking action going on. But in the race, without anyone of us modifying the cars i could not keep up with him. Even on the straights, where i used to have a higher topspeed i couldnt keep up anymore. I ended up 12 seconds behind him, just aber 3 laps in Monza.

he also wondered what happened, but i could not explain it. Later whatching the replay i realized that my breaking indicators where flashing all the time. Googeling the problem brought up some threads about ghost breaking, being an issue with the logitech wheels. I modified my pedals according to the suggestions in those threads and i never experienced such a problem again.

I dont know wether the problem stated in this thread could also be caused by the wheel, or if it is a problem related to the game, but maybe having a look at the wheel might help one or two of you.
 
I think the problem isn't really with PS3 performance, well, only inderectly. I think it's the CPU cycle priority that's ass-backwards when running GT5 online in multiplayer.

In all the race sims I've ever played, it seems that no matter what the framerate the GFX engine is running, the physics and netcode are unaffected. If the physics engine runs at 1000hz, or 100hz (or fps, same thing), the framerate you see on screen won't possible affect that because the physics engine takes PRIORITY. You could have a huge dip in GFX FPS, say down to below 15FPS, but it won't affect your laptimes or latency because the CPU cycles are there for the physics and netcode no matter what, and if the laptimes were affected, it would only be very slightly.. like hundredth of seconds.

It seems like that in GT5, PD coded it so that a hit it GFX performance can directly affect the physics engine and netcode negatively. This seems like a completely ''rookie mistake'' made by a team inexperienced in programing netcode.. EDIT- (that won't take or listen to outside help. Sound familiar?)

This also explains why the the physics are different online when netcode for multiplayer is being ''added into'' the equation. Off-line there's no netcode being run so the physics are solid. It also explains why it only happens with a lot of people in the race, the more chance GFX framerate performance issues are more probable. It also explains why it happens to people in the back of the pack, their PS3's have to render more cars (more cars in front), so more chances of GFX performance issues affecting netcode/physics.

I think what PD/Kaz have done is given the GFX framrate priority over netcode and physics engine performance in some circumstances. I'd take the GFX performance hit any day over having inconsistent physics online! It's a no-brainer.

It's another one of those WTFPD moments for me.

And I have noticed this from DAY ONE. I specifically remember GT5P having some pretty big GFX framerate drops online with a big field of cars online, more than the same amount of cars in a race offline. That FPS drop didn't affect physics though, like it can in GT5. In GT5P the netcode was correct though, which allowed for GFX framerate drop not to interfere with netcode and physics FPS. It all makes sense. I miss those GFX FPS drops in GT5P now, because online multiplayer in GT5 is FUBAR.

EDIT
To prove this theory, run 16 premium cars until problem is obvious, then everyone switch to standard cars. I bet problem disappears. All go back to premium cars... problem will magically reappear. This should prove my theory.
 
Last edited:
BWX
I think the problem isn't really with PS3 performance, well, only inderectly. I think it's the CPU cycle priority that's ass-backwards when running GT5 online in multiplayer.

...

Great post, it all makes sense (and I am a programmer, you sound like one too).
 
so can this problem be fixed with a patch or is it a problem that cannot be fixed at all? because if not then i am goin to stop playing. It has gone too far for me, its got to the stage, now i know this problem exists, that online racing in gt5 isn't possible. If PD released a patch enabling us to turn the replays on or off, would this be a fix? Im not a computor programmer so forgive me if it sounds a little noobish.
 
I'm sure it's technically fixable, but I seriously doubt PD will make any effort at all to fix it. They're far too busy coming up with the next pointless addition to the game. Racing gear socks, probably.
 
I think BWX is onto something and yeah, big rookie mistake ( maybe explains why there are no leaderboards, too)...


However, could you please at least fix the title ...? I use pad and I have this too. probably worse.


And yes GT5P online was a lot better, at least finally when they broke in and removed that horrible, horrible boost for some races.

/
 
Great post, it all makes sense (and I am a programmer, you sound like one too).
Thanks, and nope I'm not a programmer but am long time sim racer and being a beta tester for RBNET gave me a lot of insight on the subject.

Learned a lot from a guy called ''Racer_S'' who tacked on a very impressive multiplayer mode to Richard Burns Rally. That guy is a self taught hacker/programming guru/genius. =] [/OT]

http://www.tocaedit.com/IB
http://www.tocaedit.com/IB/index.php?app=downloads&showcat=2
 
Looked at some replays I have. The most recent one 3D3 racing with a grid of 14 standard cars, where I suspect I *might* have been affected. I checked about 4 of my own laps and 5-6 laps of others and _all_ of them are 3-4 seconds faster on actual stop watch compared to my GT5 record. I guess that suggests that _my_ gt5 clock was off. Gonna have to do more comparisons, perhaps with the same race record from a different driver. Two previous 3D3 races are OK, but I only looked at few of my own laps, so if anyone thinks they were affected - I can measure their times in my replays. Those races had grids under 12 though.

Another race where I was waay off my quali pace is in a mazda TC @GVS-R. Grid was 15. Again, everyone on my replay is 3-4 seconds faster on stop-watch relative to GT5 times. Plenty of people, however, were running at a pace I was consistently able to do in free-run and practice races, yet I was 4 seconds a lap slower.

Another race where I think I got affected: grid of 15 premium cars. Suzuka east. Short course. Only 1.5-2 second difference between stopwatch and gt5 clock. For everyone I checked.

Another race where I think someone else was affected (but I don't think I was): Full grid of premium Miatas @GVS-E-R. The guy in question qualified 4th, but was running about 2 seconds slower in race then the similarly-paced qualifiers. GT5 lap times (his and few of mine) are in sync.

So I take from this is that one's GT5 internal clock gets screwed up when this glitch happens. Best way to check if everyone was affected - have everyone post replays and check if GT5 clock was screwed up.
 
nealcropper
so can this problem be fixed with a patch or is it a problem that cannot be fixed at all? because if not then i am goin to stop playing. It has gone too far for me, its got to the stage, now i know this problem exists, that online racing in gt5 isn't possible. If PD released a patch enabling us to turn the replays on or off, would this be a fix? Im not a computor programmer so forgive me if it sounds a little noobish.

This.

I run the FGTC which is a fantastic racing league that has many more applicants than space. We are currently on roughly race 35 and we have never seen this problem before the update. Today at Monaco I had to abandon the race due to this slow down problem, we have been racing with 14-16 people once a week for 35 weeks with very few problems but now it seems like its finished.
My lap times fell by 7 seconds per lap, my car felt like it weighed x4 and had comfort hards on. If this isnt patched then GT5 has nothing to offer me as online is the only thing that keeps this game alive :(
 
BWX
I think the problem isn't really with PS3 performance, well, only inderectly. I think it's the CPU cycle priority that's ass-backwards when running GT5 online in multiplayer.

In all the race sims I've ever played, it seems that no matter what the framerate the GFX engine is running, the physics and netcode are unaffected. If the physics engine runs at 1000hz, or 100hz (or fps, same thing), the framerate you see on screen won't possible affect that because the physics engine takes PRIORITY. You could have a huge dip in GFX FPS, say down to below 15FPS, but it won't affect your laptimes or latency because the CPU cycles are there for the physics and netcode no matter what, and if the laptimes were affected, it would only be very slightly.. like hundredth of seconds.

It seems like that in GT5, PD coded it so that a hit it GFX performance can directly affect the physics engine and netcode negatively. This seems like a completely ''rookie mistake'' made by a team inexperienced in programing netcode.. EDIT- (that won't take or listen to outside help. Sound familiar?)

This also explains why the the physics are different online when netcode for multiplayer is being ''added into'' the equation. Off-line there's no netcode being run so the physics are solid. It also explains why it only happens with a lot of people in the race, the more chance GFX framerate performance issues are more probable. It also explains why it happens to people in the back of the pack, their PS3's have to render more cars (more cars in front), so more chances of GFX performance issues affecting netcode/physics.

I think what PD/Kaz have done is given the GFX framrate priority over netcode and physics engine performance in some circumstances. I'd take the GFX performance hit any day over having inconsistent physics online! It's a no-brainer.

It's another one of those WTFPD moments for me.

And I have noticed this from DAY ONE. I specifically remember GT5P having some pretty big GFX framerate drops online with a big field of cars online, more than the same amount of cars in a race offline. That FPS drop didn't affect physics though, like it can in GT5. In GT5P the netcode was correct though, which allowed for GFX framerate drop not to interfere with netcode and physics FPS. It all makes sense. I miss those GFX FPS drops in GT5P now, because online multiplayer in GT5 is FUBAR.

EDIT
To prove this theory, run 16 premium cars until problem is obvious, then everyone switch to standard cars. I bet problem disappears. All go back to premium cars... problem will magically reappear. This should prove my theory.

Great post,sound very logical ...

Now the question is how can we get PD to really notice this and do something about it!!!

As we all know they(pd) often check this forum so what would be the best way to get their attention?...

Maybe to change the title of this thread to something a bit more dramatic or to the point like: ONLINE RACING PHYSICS BROKEN!!!!!

Any ideas?

If nothing will happen about this issue,i am affraid many will stop racing online,and thats the last thing we want to happen!!

:nervous:



spy.
 
Get a poll thread stickied asking for input about online performance issues.

Police the thread heavily to keep the BS out.

Put an article in the news section outlining the percentage of drivers that feel the online physics performance is broken.
 
im not even sure it can be fixed, reading through the peoples posts it seems like the programers have made a desicion when writing the net code and it was a bad one. I hope it can be fixed. Realisticly, can it be fixed? There seems t b a few people involved in this thread that know what their talkin about, how much can be done via a patch, can u rewrite the core code?
 
im not even sure it can be fixed, reading through the peoples posts it seems like the programers have made a desicion when writing the net code and it was a bad one. I hope it can be fixed. Realisticly, can it be fixed? There seems t b a few people involved in this thread that know what their talkin about, how much can be done via a patch, can u rewrite the core code?

We're all guessing. They're good guesses, but without access to the source only PD will know if it's an easy fix, a hard fix or a total rewrite fix.

I would sadly assume it's not an easy fix, they must be aware about it (I made the last thread in September which ran for a bit.)
 
MGR
It's been established there's two different physics engines operating in GT5 between online and offline. There's a number of threads and plenty of evidence on this topic so I don't think anybody's debating that fact. This specific issue relates the the consistency of the online physics.



How consistent are your laps in free-run? (ie how much do your lap times vary in a 3 or 4 lap stint)

And how much did your lap times suffer because of it? (examples of lap times?)

Cheers

Sorry for the late reply. My laps are usually within 0.2-0.3 seconds of each other in free run. In the race I was roughly 3-4 seconds off (this was on Grand Valley Speedway). I can't remember what the exact laptimes were though :indiff:
 
Great post,sound very logical ...

Now the question is how can we get PD to really notice this and do something about it!!!

As we all know they(pd) often check this forum so what would be the best way to get their attention?...

Maybe to change the title of this thread to something a bit more dramatic or to the point like: ONLINE RACING PHYSICS BROKEN!!!!!

Any ideas?

If nothing will happen about this issue,i am affraid many will stop racing online,and thats the last thing we want to happen!!

:nervous:



spy.

And while we at changing titles, we should scrap the "wheel" part as it makes no sense, I use a pad and get the same 🤬.

OH and this isn't a glitch either, a glitch is something you can make happen on purpose, this is just bad netcode or simply laziness on PDs part.

So...

GT5 NETCODE BROKEN, POSSIBLE DOWNFALL, IF IT DOESN'T GET FIXED!!

maybe something like that?



_:nervous:
 
And while we at changing titles, we should scrap the "wheel" part as it makes no sense, I use a pad and get the same 🤬.

OH and this isn't a glitch either, a glitch is something you can make happen on purpose, this is just bad netcode or simply laziness on PDs part.

So...

GT5 NETCODE BROKEN, POSSIBLE DOWNFALL, IF IT DOESN'T GET FIXED!!

maybe something like that?



_:nervous:

That should do it 👍


Will send OP a quick pm...



Spy.
 
Sorry for the late reply. My laps are usually within 0.2-0.3 seconds of each other in free run. In the race I was roughly 3-4 seconds off (this was on Grand Valley Speedway). I can't remember what the exact laptimes were though :indiff:

I race exclusively online. On most days I will spend a few hours racing, been doing this since release. I've never noticed anything like this 'glitch'. My free run and race times match, in fact when I am pushing I usually lap within a 3.sec range. My lap times only vary if I make a mistake or if tyre wear is on.

In the many thousands of races I've completed I've never been 3-4 seconds off the pace in an equal car.


As for the code, net code does not take up much CPU time at all, its a marginal cost. Anyway as has been said, the frame rate drops right? WHY?

Tell me why the frame rate drops rather than the no frames being dropped and the game slowing down? If you can answer that then you know full well why the graphical processing is not going to hold up physics processing.
 
I would sadly assume it's not an easy fix, they must be aware about it (I made the last thread in September which ran for a bit.)

That was a good thread, but fell away before enough noise was raised, it seems.

What's odd about this issue is that some people claim to get no trouble and this is a non-issue, but others like me experience it close to 85% of the time.

It is frustrating, and really ruins the game. Plus, it makes the affected drivers look like complete noobs in a race.
 
So could it maybe be ps3 related in the sence of fat/slim/new-old/size of hd ??

As nasanu states i personally also never ever experienced this,but you just cannot deny all the posts about this issue......

.



Spy.
 
Two summary points, one bad and one good:

1, This issue is real.
There been some in my view some very credible posts from people who has gone through some in depth research that this issue is real.

2, Not all are affected.
Even in the same race/room where people has been affected other has not. So if we "only" can figure out what is causing it to happen, there might be a way in where we all can avoid or work around it.
 
What's odd about this issue is that some people claim to get no trouble and this is a non-issue, but others like me experience it close to 85% of the time.

That is really odd. If I were a PD developer I would get in contact with you, as you seem to have a reproducible case.

I have never had the problem myself, and I have recently been in full grid races with no problems reported, premium cars and people from different continents. So it is not like fail is guaranteed.

The mystery is when and why does hit bug hit? Perhaps it is all about probability, and a number of factors can increase the probability of fail.
 
That was a good thread, but fell away before enough noise was raised, it seems.

What's odd about this issue is that some people claim to get no trouble and this is a non-issue, but others like me experience it close to 85% of the time.

It is frustrating, and really ruins the game. Plus, it makes the affected drivers look like complete noobs in a race.

Don't get me wrong, but this might be a great opportunity for us to figure out what is causing it.

Could you please share your set-up (PS3 model, steering device, network, monitor set-up, etc). As much details as possible.
 
Back