UKGPL Season 23 (2012) Porsche Super Cup - Clermont Ferrand 65 - Sep 14

[BadBlood]

There are two main causes of disconnects - client/server mismatch and this is almost always the client as the client reacts badly to not getting an update. If it is server side then the server can lose track of the driver but usually there is not a disconnect. The other main reason for a disco is that the track has a problem.

Either way, discos caused by the server are rare.

[clouds]

There are two main causes of disconnects - client/server mismatch and this is almost always the client as the client reacts badly to not getting an update. If it is server side then the server can lose track of the driver but usually there is not a disconnect. The other main reason for a disco is that the track has a problem.

Either way, discos caused by the server are rare.

Talking about the track that in particular I like (even if I have not mastered it adequately) this is the same way of saltzburgring where I had strange behaviours with my Porsche, the worst of all was the brake bug and honestly I actually still don't know why some drivers have had this problem and others had not. It is frustrating to see that the brake bug begins with the 1st sector of the airpin (and it have to be banked) and it ends when wheels returns to the flat straight.
Definately the brake bug consist of an unwanted brake that blocks the front train leading you outside a hander along the tangent line and no matter if you brake for slowing down the car. It is like to have the hand brake pulled but functioning only on the front wheels.

[NickyIckx]

  only thing I can say ; I havnt had freezes till I upgraded my PC .
unluckily I did hardware&software wise , so hard to extract whats the prob could be .

Win7 64bit failure report says : ati fuel.services.exe caused the freeze .
But google search for that let ya get lost in universe of `could be , but who knows exactly `  

well I will check if changing back to older 12.1 driver ( 12.6. atm , which is the newest out for outdated 4870 ATI videocard ) , will have any positive effect .


... but could be an windows or hardware issue instead ... so guees I am at the very beginning of an never ending journey  

[MagicArsouille]

 ....Hi folks....
....hum ! I'm affraid it was my last Porsche race .....one more disco after 2 laps ....it's too mutch for me !!( it's make 3/3 ! ) ....it's realy sad but I can't loose any more time  for that champ....Idon't know from where come the problèm, but one thing I know ,it's append only here !!....Imake yet this saison 20 races in the french ligue and only one disco....10 races in Belgium ligue = 0 disco....last sunday I make the 3 h of kyalami on HSO, and I brock the engine after 2h30 racing whith the Porsche ....in the UK F2 Champ s23, I make 2 races = 2 disco ......Porsche cup , 3 starts = 3 disco !!......at the moment , the 65 mods  work very good , I got 4 starts ....and 4 finishs  .....I hope this good serie will continu .......
...I will try ( last time ) again the F2 at Silver tomorrow ......

....see you guys 

[BadBlood]

....Hi folks....
....hum ! I'm affraid it was my last Porsche race .....one more disco after 2 laps ....it's too mutch for me !!( it's make 3/3 ! ) ....it's realy sad but I can't loose any more time  for that champ....Idon't know from where come the problèm, but one thing I know ,it's append only here !!....Imake yet this saison 20 races in the french ligue and only one disco....10 races in Belgium ligue = 0 disco....last sunday I make the 3 h of kyalami on HSO, and I brock the engine after 2h30 racing whith the Porsche ....in the UK F2 Champ s23, I make 2 races = 2 disco ......Porsche cup , 3 starts = 3 disco !!......at the moment , the 65 mods  work very good , I got 4 starts ....and 4 finishs  .....I hope this good serie will continu .......
...I will try ( last time ) again the F2 at Silver tomorrow ......

....see you guys 

Strange. The server you run the 65's on is the same as the Porsche. I wonder if there is a problem to do with Friday evenings (scheduled maintenance or something) as both the F2 and Porsche are Fridays.

Shame though.

[Roberto Teso]

Strange. The server you run the 65's on is the same as the Porsche. I wonder if there is a problem to do with Friday evenings (scheduled maintenance or something) as both the F2 and Porsche are Fridays...

If I can give my humble suggestion, it is not said that a server that work well with F2 or 67 or 69 have to be the same for GT or, in near future, for Can-Am.
The latter two mod are very demanding in terms of frame rate due to high polygons number and geometric complexity of cars models.
The way GPL use to exchange net info in its client-server model, it is quite credible that decreasing fps could cause degeneration of the connection process.
In past time similar problems had arisen even for a too large timing differences between clients, in that case the too large difference between delays results on large fps drop and/or discos.
I know that this could seem strange but one of the best tech in GPL was it's net-code and the way it was tied to the client-server process, same time this is a real bottleneck in many cases.

[BadBlood]

Understood but the server doesn't have to render all the graphics. It simply keeps track of the relative positions and returns those to the client. My UKGPL3 server is extremely poor in terms of its graphic capabilities but has run many GT races flawlessly. The most critical factor in all of this is the quality of the connection between client and server and as more and more of the internet gets bogged down with YouTube that connection is vulnerable. If a server is in London and the clients are in Italy, France and the UK it is entirely possible for the UK clients to be rock solid and have the French/Italian clients diconnect. The quality of connection between computers remains the biggest impediment

[Roberto Teso]

Pardon me BadBlood, but there's no connection between a pc graphics capabilities and fps in GPL, poligons in GPL depends only on CPU and the relation between this and net code is due to the 1 / (fps) comunication rate hardcoded into GPL, so, not only car models are related to but even on track polygons and net efficiency.

[Cookie]

Roberto this PC I use as an exclusive server is an Athlon ll X2 250 @ 3000 MHz and 4GB RAM.
It has no problems with GT mod serving even in 60fps -> Dec 27th 2011 it ran the Ferrari Challenge with 16x 330/P4 without problem!

IMO its more related to track and connection issues

[Hristo Itchov]

Pardon me BadBlood, but there's no connection between a pc graphics capabilities and fps in GPL, poligons in GPL depends only on CPU and the relation between this and net code is due to the 1 / (fps) comunication rate hardcoded into GPL, so, not only car models are related to but even on track polygons and net efficiency.

Of course the graphics card makes a difference! Try it with a slower card without changing the cpu and you'll see. Even though CPU matters more in GPL than other games, it doesn't mean it's all that matters. GPL still uses hardware video acceleration via its D3D/OpenGL rasterizer, at DX6-DX7 level.

[clouds]

Is there any low res Clermont Ferrand track to try ? I'm starting to think there are some tracks that are dangerous for some of us.

[Cookie]

Have you tried Ginettos fix? ->http://srmz.net/index.php?showtopic=2734

[Roberto Teso]

I don't say that GPL don't use graphics card, I've said that polygons are calculated from CPU, I know well that polys are filled by graphics card and this could arise other kind of problems.

What I want to focus is that in GPL:

1. Polys calc is tied to CPU.
2. Polys fill are tied to GPU.
3. Frame rate are tied to polys draw.
4. Net code are tied to frame rate.

So all are tied one to the others and when one hold back all the other suffer and is well known that GT and now Can-Am is very polys intensive.

Other trouble could happen using some modern, related to the age of GPL, graphics driver, DX7 (the latest DX usable by GPL with D3D patch) is emulated or wrapped by graph drivers, and this could be an hard wall hit.
It's well know that ATI had (and have) some problem to build a good OpenGL driver, same way Nvidia have similar issue in develop D3D driver.

[Hristo Itchov]

I don't say that GPL don't use graphics card, I've said that polygons are calculated from CPU, I know well that polys are filled by graphics card and this could arise other kind of problems.

What I want to focus is that in GPL:

1. Polys calc is tied to CPU.
2. Polys fill are tied to GPU.
3. Frame rate are tied to polys draw.
4. Net code are tied to frame rate.

So all are tied one to the others and when one hold back all the other suffer and is well known that GT and now Can-Am is very polys intensive.

Other trouble could happen using some modern, related to the age of GPL, graphics driver, DX7 (the latest DX usable by GPL with D3D patch) is emulated or wrapped by graph drivers, and this could be an hard wall hit.
It's well know that ATI had (and have) some problem to build a good OpenGL driver, same way Nvidia have similar issue in develop D3D driver.

I guess you understand it much better than me then. 

It really depends on the PC configuration. I haven't had screen freeze problems with this machine since I bought it, and it's not that powerful of a PC, I still have to reduce some details for the heavier mods like GT or 69x. Apart from tracks that are well know to cause screen freeze, it's been OK for more than a year that I've been using it.

[BadBlood]

Do you know for a fact that Net code is tied to frame rate? I have not disconnected when I have had frame rates of 6! The server will send out the updates at 36/sec regardless of the client frame rate. The conversation with the server is based on the Client net-send. If this is tied to frame rate (which I am unsure of - bad decision if so) then the server may well disconnect the client because of a lack of timely response but I am fairly sure that the biggest component in this is the pipe between client and server (e.g. Internet) and this pipe can and does fail.