Quote from Shotglass :

from what he wrote hes only struggling when a lot of cars are around
the car models dont have nearly enough polys to cause a drop as large as the one you experience in lfs on grounds of gpu limitation (100+ => 30 in my case)

He also said it doesn't max out his CPU (or at least one of the cores), which points in the opposite direction.

But you're right that a full grid will shift the load from GPU limited to CPU limited in most cases. This is most likely because LFS doesn't use the API in a way that is efficient on modern hardware, so it spends too much time making API calls to keep up with the GPU. I'm certain LFS could see a massive improvement by just changing to a newer API and taking more care in batching to limit state changes. You don't even need multi threading to make that happen.

Quote from Forbin :

So to summarize:
CPU = Athlon XP 3000+ (2.1GHz)
LFS average framerate: 4600 (no AA) = 6800GT (no AA) = 6800GT (4xAA)

Obviously the minimum framerate was also the same because the CPU didn't change.

Absolutely, but on my (old) Athlon XP 3000+ with a 7600GT running with 8X AA at 1920x1200 caused a significant FPS drop meaning it was GPU limited. Adding more CPU power at that point won't have much of an effect.

Boris has a 8600GT which is pretty much the same as a 7600GT (despite the higher number) and is using a resolution very similar to mine, hence it's very likely he's GPU limited, especially since his CPU is significantly better. The only way to find out is to lower the AA level, and see if that helps.

Quote from Boris Lozac :

All i'm saying is, "god forbid" that we have advancier graphics, like some more polygons on the cars for starters, pixel shaders, rain, what system would you have to have in order to work fine on start with ~15, 20 cars?

That's the ting. You could easily add very advanced pixel shaders to LFS' current graphics without taking ANY hit in FPS. The way LFS is currently using the GPU the actual shader pipeline is not really doing much work and on modern cards there is a hell of a lot of headroom there. The bottleneck is elsewhere on the graphics card (most likely memory bandwidth) and that doesn't have to change when adding more realistic shaders.

Once you have a bottleneck somewhere in a system like this, increasing the workload on any other part of the system won't really have any effect on performance. The GPU isn't some black box with finite performance. Depending on the workload different parts of the GPU may be the thing holding it back.

Quote from Gunn :

you don't need to run that high to enjoy LFS.

No, of course not . But the point is that it's quite easy to bog down a 7600GT/8600GT in LFS if you up the resolution and AA/AF, and I'm certain that's the problem in Boris' case.

Quote from Gunn :

I'm using a 7600GT and it handles LFS with its ram chips tied behind its back.

With 8X AA? Mine sure didn't. (This is at 1920x1200 BTW. At lower resolutions it might handle it.)

Quote from Boris Lozac :

The way LFS curently looks(it looks real, but the actual graphics are from '98) and the fact that it doesn't even use my CPU 100% shows that there's something wrong there.

No, like Jakg said that shows you that the GPU is the bottleneck in your system and it is holding the CPU back. Speaking from experience with a 7600GT (which in many cases is just as fast as the 8600GT), 8X AA is just too much, even in LFS. Try going own to 4X and see if that helps. If it does, the GPU was the bottleneck, plain and simple.

LFS is surprisingly hard on the GPU with it's settings at full tilt. Don't let the fact that it's "DX8" fool you. The shader pipeline on the card may very well be more or less idling when LFS is running, but there are other things on the GPU that can become a bottleneck in these cases (texture fetch, triangle setup, tessellation etc.). It's not as simple as "GPU power".

Last edited by wien, Sat 29 Dec 2007, 14:56. Reason : tessellation != rasterization :P

Quote from Bob Annesley :

So why is the BF1 etc not like this if we are to follow the concept that this patch is more realistic?

Do you people even try the patch before coming here complaining? The BF1, FO8, FOX, XRR and FXR all have ignition cut, meaning you don't have to clutch (or lift) when shifting. You only use the clutch to launch it off the line. In the BF1 you don't even need it there.

Last edited by wien, Sat 29 Dec 2007, 06:25.

I'm a freelance web-developer, which pretty much makes me a bum too I guess. Latest creation worth mentioning: http://www.kjøkken-elverum.no.

Last edited by wien, Fri 28 Dec 2007, 23:31.

Quote from JTbo :

See how nice songs here is:
http://modarchive.org/index.php?request=home-chart

Holy crap, that was a trip down memory lane. So many songs I remember listening to way back when, and there they all are.

Quote from pb32000 :

If there's no encryption anybody can just jump in and use the connection, however simple WEP is likely to stop the majority of the population from stealing wireless network traffic.

I actually had a bastard stealing my bandwidth within minutes of activating my wireless router after I bought it (WEP turned on). And I even live in a sparsely populated area. I'm not saying that is normal, but I was shocked at how easily the guy got in.

Nope. DFP and button clutch in patch Y. Used auto-clutch before though.

Quote from ajp71 :

I don't know if I've ever read anything officially confirming nK having chassis flex but people seem to think it does.

Yeah, I spent a few minutes searching RSC and I couldn't find anything but a few people thinking they saw chassis flex. Maybe that's where I got it from.

Anyway, I should think an approximation of sorts could do the job pretty well in LFS. I don't see the need to go directly to a complex model like in RoR and spend 3/4 of the processor budget on the chassis. The tyres are already very rough approximations, and they work fairly well.

Quote from ajp71 :

RoR shows the limits of current computing power, one or two vehicles in a very low detail system running at an uncomfortably low frequency can grind todays top notch machinery to a halt with very little in terms of non-chassis physics demands.

Hmm, correct me if I'm wrong but doesn't nKPro have chassis flex? Even visible chassis flex? I'm certain I've seen a slow-mo video showing it off. I'm not sure how advanced it is and how much of an impact it has but if nKPro already has it it can't be that processor intensitive? High frequency, sure, but it can probably be simplified a great deal and still give approximately the right results. I doubt a physics model like in RoR would be the only way to implement this.

FSVS is vsync. But with some drivers vsync doesn't get enabled if it's set to "off" in the drives. In the ATI drivers for example, you need to set it to "off, unless application specifies" for the in-game setting to work. I think Nvidia has a similar setting.

Quote from J. andersson :

I payed to play THE Q PATCH back...

Then feel free to play the Q patch until your arms fall off. Nothing's preventing you. Furthermore you paid for a program with the following in it's EULA:

Quote from Live for Speed End-user license agreement :

1.4 You must be aware that we can alter any aspect of LFS as we see fit. Improvements, fixes and/or changes made to the game, are to be expected.

Not helpful really, but like romus74 I'm on a 3850 (which is basically a slower 3870) and I had no problems on neither the 7.11 nor the 7.12 drivers (Vista 64). AA & AF also works just fine if I set it in the control center, both manually and through profiles.

Have you tried completely removing the ATI drivers and installing them again? You could also try updating motherboard drivers. Could be an issue there I guess?

Also, you said it only happened in driving games? Sure it isn't related you your wheel/controller somehow? Have you tired disconnecting it to see?

Unless you have set vsync to "always off" in the driver control panel, "Full Screen Vertical Sync" should enable vsync. Minimum sleep is the minimum time LFS will sleep between each frame when the framerate is locked to a specific frequency. This can be use to give the OS "time to breathe" if you have problems with sound stuttering or network lag or stuff like that.

Quote from joeynuggetz :

I'll try locking my frames at 100 FPS and try to ignore the tearing to see if thats it.

Unless you have a CRT monitor capable of 100Hz refresh rates, that won't do anything. Your screen will probably redraw at 60Hz anyway, negating any benefit from upping the framerate (It may get a little bit smoother, but the fundamental problem is still here).

Also, the cars vibrating wouldn't have any impact on the smoothness of the TV-camera which you said also was jerky.

Depends on the car. The race cars only have slicks available as they don't really have the ground clearance to do rally cross.

Worth a try anyway. I think joysticks may be more susceptible to this problem as they have all that weight swinging on top as opposed to wheels, so you may need to go a bit lower because of that. For the record I only use about 40% on my wheel, and I don't see any oscillation.

Quote from J.B. :

There was a video in the other thread that shows the irregular framesteps from the 100 Hz physics to 60 Hz screen conversion.

Yep. If you sample a 100Hz physics loop at 60Hz, the 3rd and 5th frame of every 5 physics frames will never be rendered to screen. (This varies slightly of course since this is never exact) These dropped frames results in slightly jumpy animation.

You can see a similar effect if you watch a movie (24FPS) on TV or DVD (50Hz or 60Hz depending on where you are in the world).

Quote from Fat-Alfie :

But it's NOT real life - it's just a computer program :\

A program who's goal is to simulate real life as close as practically possible.

Try turning down the FFB strength inside LFS. This usually happens if you use too high FFB strength.

Quote from tristancliffe :

To be honest, the latter is also useless without video and or datalogging, as I can say I've slipped a clutch by 50% at 19000rpm for 829 minutes, and you have no evidence either way (I didn't by the way).

Well, yes. But once you throw your results out there they can easily confirmed/disproved by other people with similar experience. Cold hard data obtained through rigorous testing is always best of course, but since good quality data is usually hard to come by in this field anecdotal evidence isn't completely worthless. If a lot of people with experience agree, you're probably on to something.

Anyway the point was, anything other than "I don't like it because it's different". Back up your claims and you'll usually find people listen to you.

Quote from bbman :

Nope, they whined over a change in one of their beloved cars (how dares Scawen touching it!) without even thinking of adjusting the gears or camber, as Scawen urged so many times...

Yep, and that was my main gripe with that thread as well. There is only "I don't like the new clutch" instead of "I don't like the new clutch as it differs wildly from the clutch in my real life car. I can slip the clutch at 3000 revs for 3 minutes in my real car before overheating, but in LFS I can only hold it for 30 seconds before doing the same." See what I'm saying? One is a completely subjective opinion and has absolutely no merit, the other is an objective fact that can be easily confirmed/debunked by a third party.

We want the latter type of comment. The first is worthless. If you want something changed, be prepared to back up your opinion with some facts.