Update 1.02 DID NOT fix the input lag [UPDATE 5]

iLex
We should not forget however, that the whole input lag issue came into this world because people were wondering how on earth the cars could be this terrible to drive. There were people who stopped playing...
Click to expand...

Your correct, "some" have stopped playing, many others not, but it is really just based on personal preference. S2U doesn't drive "terrible" - in the "opinion" of many!
Just because I haven't picked up GT5 since S2U was released doesn't mean GT5 is a bad game - no, just my personal preference. I totally enjoy what S2U has to offer - as many others have also stated in the fourms.
All of the racing games/sims I play have a different feel. I don't expect them all to be the same - and thankfully they all are not the same.

This whole lag thing is getting so old!
 
TheTechnobear
My point was the quoted text, tries to claim superiority based on a wild assumption which is incorrect.
Click to expand...

Ah, ok. Sorry, I didn't read any superiority in that comment, but that might have been me.

TheTechnobear
Yes, 1.0, I found nearly unplayable on ps3 ( xbox was better). This was widely reported...
but IMHO if the community determines it's input latency or not is irrelevant - report the problem (handling) and leave it to the devs to determine the cause(s). They have much better tools to determine the underlying issue and also know the code base - users just use pure speculation on how it might be implemented based on a healthy dose of assumptions.
Click to expand...

I agree that it is ultimately something the devs should examine and fix. However, I do like the fact that the community analyses these problems. The whole community has quite a knowledge of video games as it doesn't only consist of people who play games, but also people who (have) work(ed) in the field. Granted, overanalasysis might spoil the fun, but a bit of a discussion on these topics can in my opinion only be for the good.
 
bmwisme
Your correct, "some" have stopped playing, many others not, but it is really just based on personal preference. S2U doesn't drive "terrible" - in the "opinion" of many!
Just because I haven't picked up GT5 since S2U was released doesn't mean GT5 is a bad game - no, just my personal preference. I totally enjoy what S2U has to offer - as many others have also stated in the fourms.
All of the racing games/sims I play have a different feel. I don't expect them all to be the same - and thankfully they all are not the same.

This whole lag thing is getting so old!
Click to expand...

I was ofcourse referring to the time before patch 1.02 when pretty much everyone who had ever played another racing game thought the handling was at least a bit odd.

There is a difference between not playing a game because you like another one better and not playing a game because it is not what you expected. Both can happen at the same time, though.

I too completely enjoy S2U and haven't touched GT5 since I got this game. However, this doesn't mean the game doesn't have flaws/bugs/things that need fixing.

And I don't expect all games to be the same, I do expect them all to be of the same standard.



edit: Sorry for the double post.
 
iLex
...Granted, overanalasysis might spoil the fun, but a bit of a discussion on these topics can in my opinion only be for the good.
Click to expand...

Yes, but the lag issue has been beaten to death! I have heard many in other forums, many who haven't even played S2U, state how bad it is - based on second hand knowledge.
The GT5 forum has no where near the rants as this forum - am I just that wrong in my assessment - maybe I am.

Look at the GT5 forum topics:

GT5 Questions & Answers
GT5 Seasonal Events
GT5 Marketplace
GT5 Photomode Forums
GT5 Tuning Forum
GT5 Course Maker
GT5 B-Spec Forum
GT5 Online Racing
GT5 Videos
GT5 Race Reports
GT5 Drifting Forum

I find very little focus on the GT5 shortcomings in these. No one seems to mind gaining track postion by bouncing off the other cars, huge air on jumps with smooth landings, very little loss of grip on runoffs, winning every race, the horrid menu system, etc..

I am at a loss why this is so. My $.02 - not right or wrong.
 
iLex, I didn't mean to single you out - hope you didn't take it that way. I just used your coments as a start for my post on the lag rants. Take care...
 
bmwisme
iLex, I didn't mean to single you out - hope you didn't take it that way. I just used your coments as a start for my post on the lag rants. Take care...
Click to expand...

No worries, didn't take it that way at all. :)

You are ofcourse right that the issue has been beaten to death.

DrJustice
And you're doing your bit to perpetuate the discussion :sly:

But yes, it's old. And tedious. It's time for a final patch so that it don't get much older :lol:

DJ
--
Click to expand...

And this is why we all keep the issue alive... in hopes of a fix.

Regardless I agree with the statement that the game (1.02) is a lot of fun and it's time for the PS store to come back online so PS3 users can buy some DLC!
 
bmwisme
The GT5 forum has no where near the rants as this forum - am I just that wrong in my assessment - maybe I am.
Click to expand...

Oh, I think in fairness when gt5 came out, it had it's fair share of rants.
( std vs premium, grinding, why it took 5 years etc etc)

Seems to be the norm with forums, a game gets released, the rants start ... then after awhile those that enjoy the game stay, those that don't (presumably) move on.
Fm3, f1 2010, gt5 all had it (so did tdu2 but that was self induced :)) ... Dirt 3 seems to be fairing okay at the moment.


Talking of patches, apparently 1.03 is ready for submission to EA, then presumably Sony/ms
(source griff @ no grip)
Likely to be a few weeks I guess, and there seems to be no info as to what will be in it.
 
TheTechnobear
My point was the quoted text, tries to claim superiority based on a wild assumption which is incorrect.
Click to expand...

Are you serious? How on earth could my statement be a claim of superiority? I merely pointed out that some folks are not sensitive to the lag under discussion. That seems like a very fair assessment and not at all a criticism.

You seem very anxious to minimize this issue. For what reason, I can't begin to guess. If you are not having a problem with lag then go ahead and enjoy the game. (I am). Just please stop assuming (and yes - YOU are the one making an assumption) that by pointing out the obvious I'm trying to claim some bizarre kind of superiority over you. Get over yourself.
 
Last edited:
DrJustice
And you're doing your bit to perpetuate the discussion :sly:
...
--
Click to expand...

Your right... I could just avoid the topic - just never know if I could pick up a tip or not so I find myself scanning.

It's good to know there is talk of another patch. The more "happy" campers the better!
 
boxox
FWIW here's what I get:



(sorry for the delay, nobody wants to see my gut looming into the camera :P)

This is a fairly old proview 19" lcd and TBH beyond it's contribution to things I'm kind of struggling to see what the issue is here :)
Click to expand...


DROOOOOLLLLLL....... :)

That's just 0.133 seconds. Man I would love to have that on PS3. I might have to start giving some serious thought to connecting my PC to my TV!!! Very nice.
 
Ok, so, from the fingers of Griff:

Some comments about latency and measuring it:

- LCD/LED monitors have latency - most have a 'gaming' mode that turns off any filtering or processing which is a must when recording a video, so that the frame buffer 'in the tv' hasn't been multiply buffered. There is then a response time which is a quite subjective, debated area - it certainly isn't zero and it's not my place to argue claims on 2ms latency etc. I tend to add 10-20ms realistically.

- Driver frame 'pre-rendering' - to smooth frame-rate most driver vendors can queue upto 3 frames by default to somewhat hide frame to frame variation when a game submits the scene. The interaction here is largely out of our control - Shift 2 specifies two back buffers but the exact frame reveal timing with the driver isn't defined by us on a present d3d call.

- Simulation versus visual feedbaack : our game processes the inputs on a separate thread and these are feed to the physics. There is a snapshot that is then used by the rendering system - this is called 'bridging', because the communication between the renderer and sim goes through an intermediate interpolation step, which uses 'system' time to move all aspects of the scene perfectly stepped against when the simulation time was executed.
This is quite a common practice to allow multi-threading. It does add latency - so the renderer is always one simulation step behind in it's rendering.

- Frame rate : because of the above latency components the total lag can be proportional to the individual frame ms time as a function of simulation lag, driver buffering and display lag, the net result being that lag/latency can increase at lower frame rates.

- Wheel and tyre position simulation : recording a video of an in-car wheel against a real world input is furthur distorted, since the animation system uses multiple frame smoothing. So there can be additional latency on this rendering - this doesn't affect gameplay, unless the wheel(s) is/are your focus. I can see this might be distracting and it's hotly debated amongst devs how to balance visual feedback against the underlying simulation.

Currently my own estimate is 90-140ms latency at 60fps for our game.
Click to expand...
 
I would like to say that would be right in this topic speak about just for the ps3 version of the game; I have both the pc version and ps3 version of shift 2 and result after the patch is completely different.
On my PC (core2duo 8500, ATI 4870, 4 gb ram) before patch I had the same amount of imput lag of PS3 version (prepatch).
After patch 1.0.1 (the only one patch for pc version) the lag is almost disappear (as the video of boxox) but the things are very different on PS3 after the patch 1.2 (the second one) where people more sensitive to the lag can still feel it and complain about it.
 
boxox
Ok, so, from the fingers of Griff:
"...Currently my own estimate is 90-140ms latency at 60fps for our game."
Click to expand...
Thanks for the reference, boxox. Of course that description is for the PC platform. Pretty much in tune with my previous speculation, including 2 queued back buffers, but with some extras added (d3d/driver-vendor issues and 'bridging' of inputs). Things may be handled differently on consoles though, I'd not be surprised if they're more efficient, e.g. in that there is no d3d/driver-vendor issues and things probably run leaner in general. I have not seen the PS3 SDK, but my impression is that you're pretty close to the metal on that platform.

Just for fun lets take Griffs process description and median latency number (110ms) and run with it in a console setting. For simplicity, let's say that the latency contributed by queued display frames is due to 4 pipelined frames on average. Now we substitute the 60Hz framerate with th 30Hz framerate and get 176ms. Then add some for the remaining processing on the weaker platform and subtract some for it's (probable) higher efficiency, and a reasonable worst case number might lie below 200ms. That number could (ought to...), again technically, be a fair bit lower than that.

As usual this is speculation, but I reckon it's not unreasonable. At least it's fun :)

DJ
--
 
Very likely it's going to be (ideally - if it's working :P) the standard ~100ms/133ms range (excluding i/o / monitor refresh) for a multithreaded double buffered console game at 30fps.
 
Last edited:
All very interesting stuff but it just does not correspond to the PS3 experience at all. As I mentioned earlier, my choice to use the tire reaction as a measuring point vs wheel input was to take suspension loading out of the loop. If I measure my wheel input vs the first discernible direction change of the forward view from the car that adds approx 70ms to the measured 366ms for a total of 436ms (approx). This applies to hood cam, cockpit view, whatever. All of those tests were done with my TV on gaming mode with every possible video processing option turned off. So while it is interesting to read these comments from Griff it is clear that either he is directly addressing the PC platform experience or he simply is not paying any attention to the console differences. I'm beginning to suspect the latter unfortunately. Any explanation from him indicating that the additional lag on PS3 is due to matters beyond their control is a total cop out. GT5 : 270ms, S2U : 436 ms. 'Nuff said.
 
Last edited by a moderator:
@drjustuce , I think SMS have implied the ps3 does have some difference ( see digital foundry interview) due to the fact that certain tasks cannot be as placed on the spu, and so places more load on the single CPU of the ps3.
Of course, you can only speculate on what this means... e.g. Is input io more likely to be competing for resources with physics engine ( completely fake example as I don't have acces to source )

@pgag,
have you looked at the times griff posts, usually midnight his time (gmt) , ie he does it in his own time for enjoyment/pride.
As a developer as well, I can understand him getting a bit fed up sometimes as people misunderstand what they can expect of him.
Also, he does not have a customer facing role - for that you can go to EA, if you want a politically correct response.
Finally, he is the pc lead, so why do you think he doesn't know about the pc? Like any member of a team, he will not know all details, but I'm sure he has an enormous amount of interesting information to share - should he feel like it!
 
TheTechnobear
@pgag,
have you looked at the times griff posts, usually midnight his time (gmt) , ie he does it in his own time for enjoyment/pride.
As a developer as well, I can understand him getting a bit fed up sometimes as people misunderstand what they can expect of him.
Also, he does not have a customer facing role - for that you can go to EA, if you want a politically correct response.
Finally, he is the pc lead, so why do you think he doesn't know about the pc? Like any member of a team, he will not know all details, but I'm sure he has an enormous amount of interesting information to share - should he feel like it!
Click to expand...

I completely understand that he is going above and beyond the call of duty by participating in that forum and I applaud him for it. I also totally understand him getting fed up with some of the inane and ridiculous comments he received. However, as for him not having a customer facing role - that is completely incorrect. Whether it is part of his job description or not he has chosen to take on the most visible customer facing role there is - by directly interacting with customers on a public forum. As such he should conduct himself in a professional manner befitting that which would (or at least should) be expected of him while representing his company. In that regard he has completely failed.

Regarding his knowledge level of the PC product - you are correct and in retrospect I should not have posted anything derogatory in that regard. I based my comment on my observations from that thread where he seemingly replied with great enthusiasm to anything graphics related but frequently ignored other topics. As you rightly point out however no one guy can know it all and there are likely other forces at play (such as EA) that may prevent him discussing certain topics.
 
Nah the input lag has not been fixed.To get the primary part of a racing game this wrong is a joke but some people on this forum would be happy to play this game with playstation move.As long as you have your stickers and bonny colours its fine. The driving part comes last. What a joke:tdown:
 
pgagoober
Oh never mind. No use arguing with a fool. Back on topic.
Click to expand...

Boxbox can be many things but I dont think he is a fool and just because you two dont see eye to eye theres no use in calling him names...
From were I see it Griff isnt perfect (nobody is) but I appreciate what he is doing for the PC comunity very much and I just wish that the guys responsible for console coding would do the same.... but why should they... if you get critized whatever you do, so they might just as well do nothing and leave us all in the dark...
 
He was probably out of context and didnt notice the 1.02 in the title wich obviously means the OP was talking about the consoles games... or just wanted to show it was a console issue only... it is an issue nevertheless but not as bad as it was before the 1.02 IMO...
 
boxox
How dare I post about how it's a silly idea to use the wheel animation as a measure of lag and then relay Griff's post that it's a dumb idea to use wheel animation to measure lag?
Click to expand...

I'm sorry pgagoober but boxox is right. After reading some of his posts, it seems to me that there may not be a relationship between the animation of the steering system and the actual steering of the car while it is in motion (a theory which I have wrongly ignored in my OP). What pgagoober should do is test the lag between the real steering wheel and the steering of the car in motion (which he exactly did before the 1.02 patch), and then either give us the results in video or in numbers. Only then can we REALLY tell how much of the input lag--the real input lag--was fixed.

And as for this off-topic argument over Griff's comments, one of you may as well create a separate thread on it before the said argument turns into a violent flamewar and derails this whole thread to lockdown.
 
Not to worry, the Griff sideshow is over. My apologies for the detour ;)

I agree measuring the reaction from the wheels is not very accurate and perhaps not even directly related to the perceived lag in vehicle direction change. The first videos I did on the prepatch game focussed on the observable direction change after input. That measurement was consistently 70ms longer than the wheel measurement so there appeared to be a relationship, even if not a direct one. The goal of measuring the wheel reaction was to eliminate any possible suspension loading delay from the physics engine. It seemed unfair to include a potentially accurate and realistic delay in the overall lag measurements.

I will repeat that same direction change test as soon as I get a chance. I expect the results to be somewhere around 430ms. If that proves to be correct, then patch 1.02 reduced the lag by about 70ms. It doesn't sound like much but even that reduction made a very noticeable difference in playability. I gotta wonder how much further improvement would render the lag effectively unnoticeable. I suspect it won't take much.
 
pgagoober
Not to worry, the Griff sideshow is over. My apologies for the detour ;)

I agree measuring the reaction from the wheels is not very accurate and perhaps not even directly related to the perceived lag in vehicle direction change. The first videos I did on the prepatch game focussed on the observable direction change after input. That measurement was consistently 70ms longer than the wheel measurement so there appeared to be a relationship, even if not a direct one. The goal of measuring the wheel reaction was to eliminate any possible suspension loading delay from the physics engine. It seemed unfair to include a potentially accurate and realistic delay in the overall lag measurements.

I will repeat that same direction change test as soon as I get a chance. I expect the results to be somewhere around 430ms. If that proves to be correct, then patch 1.02 reduced the lag by about 70ms. It doesn't sound like much but even that reduction made a very noticeable difference in playability. I gotta wonder how much further improvement would render the lag effectively unnoticeable. I suspect it won't take much.
Click to expand...

Why got to all the trouble of measuring changes of direction instead of simply measuring the brake lights? It has a nice, sharp transition from on to off, so you eliminate the "is it turning yet" factor.
 
Imari
Why got to all the trouble of measuring changes of direction instead of simply measuring the brake lights? It has a nice, sharp transition from on to off, so you eliminate the "is it turning yet" factor.
Click to expand...

Hi Imari. I guess because the brake light test can have the same doubt cast on it as the tire test. It is not directly related to the direction change after input, which is the primary problem.
 
^ It's also possible that the gas and brake are processed differently from the steering input. We now know that the latter undergoes several stages of processing/filtering: the steering itself, the steering wheel animation and the front wheel animation.

I reckon that testing using the cars directional change as a measure should be pretty accurate, using a low enough speed so that tyre slip isn't a big factor and a sudden large steering input, then subtracting a nominal time (a few tenths of milli seconds?) for the wheel input to be large enough to cause a significant (observable) directional change. Of course, devising an unquestionable test protocol is always a challenge.

I keep wondering, what is the idea behind not having a 'rigid' connection between the players steering wheel, the animated steering wheel, the front wheel animation and the steering input to the physics engine. I.e. just a plain simulation of the actual steering components :confused:

DJ
--
 
Last edited:

Similar threads

LM7325
NFS Shift 2, a 2023 PS3 experience in progress.
Replies
5
Views
6K
Mercury1
Mercury1
zim988
HIDUSBF + ANY SIM RACING WHEEL
Replies
0
Views
319
zim988
zim988
Tool1312
Incoming v1.4 update/patch
Replies
0
Views
2K
Tool1312
Tool1312
TagsxA
SteamVR with iRacing "lag"
Replies
0
Views
2K
TagsxA
TagsxA
M
Lag or timing reset?
Replies
1
Views
494
mattikake
M

Latest Posts

Back