I started the race for our team in SO4R in GTAL 2010, starting about midpack and don't recall any slowdowns. I will test again this evening though just to be sure (and also watch from last on the grid).

The Q6600 is quite literally two E6600 cores on the same die. The Q6600 has 2x4MB of L2 cache but each core can only access its associated 4MB, so performance there will be exactly the same as an E6600 under ideal conditions. I think the Q6600 will have a slight advantage of background tasks are being computed on one of the cores not sharing the 4MB with the LFS core, but that's it.

That review by Eza is really nice, and does look like I am mistaken in theory. In most of his tests (from 2.5 year old computers) the minimum fps was about ~55 (low end) and 70-80fps (high end, of 2.5 year old computers). Hardly any need for multithreading there.

When I used to race back in the CTRA days, would have a full field of 40 cars and on the maximum graphics, 4x AA/8x AF at 1280x1024 would not drop below 60fps.

Q6600 stock, 2GB RAM, ATI 2900HD.

So in other words, you suck.

Quote from pik_d :

That review by Eza is really nice, and does look like I am mistaken in theory. In most of his tests (from 2.5 year old computers) the minimum fps was about ~55 (low end) and 70-80fps (high end, of 2.5 year old computers). Hardly any need for multithreading there.

IIRC, Scawen or Victor once said that if we all had monster CPUs, the patch would have been out. I assume this also accounts for multithreading, because not having enough ST performance and not looking into MT in an application that should scale well with cores, is doing it wrong, very wrong...

And with an outdated DX8 graphics with no graphics updates in site, the newer IGPs from both Intel and AMD won't be a bottleneck.


@S14 DRIFT
Max is 32 cars on the track.
1280x1024 is quite a low resolution nowadays, maybe his monitor is bigger...

yeah I realise that I meant 40 on server with 32 on track. My bad, its omgwtfAM!

But you get the point. And even at 1440x900 (which is what I've gone to) I've only noticed a 3-5 fps drop in games. (not LFS as I dont play that)

Quote from AutoPilot :

IIRC, Scawen or Victor once said that if we all had monster CPUs, the patch would have been out. I assume this also accounts for multithreading, because not having enough ST performance and not looking into MT in an application that should scale well with cores, is doing it wrong, very wrong...

And with an outdated DX8 graphics with no graphics updates in site, the newer IGPs from both Intel and AMD won't be a bottleneck.


@S14 DRIFT
Max is 32 cars on the track.
1280x1024 is quite a low resolution nowadays, maybe his monitor is bigger...

That may be ST vs. MT or it may just be needing to aggressively optimize for lower end computers regardless of thread count.

Just because something is DX8 doesn't mean integrated graphics is powerful enough to handle it. There's many cars that support a certain DX# but cry when they try to run a game with it.

And check out the link in Gener_AL (UK)'s post, it shows that resolution doesn't change FPS with a graphics card about as good as S14's.

I saw that link when Eza first posted it (I've even posted in that thread if you go through it ). I'd say that 4870 is a fair bit faster than ATI 2900, but it's true that LFS is largely CPU limited.
The numbers there are based on the current patch though, unless you think that physics is OK as it is, I still disagree that MT isn't needed. You can always get a bit more refined and more complex model with more CPU performance. And LFS being singlethreaded, there's a LOT more headroom going MT than trying to squeeze more from ST.

The only thread about what I said that I could find in my brief search was this: http://www.lfsforum.net/showthread.php?p=1426006#post1426006

I didn't say it was just DX8, I said it's also outdated, which is true. I mean it ran pretty fine with my Ti4200 back in the day without too many cars around, and IGPs on Sandy Bridge, and especially the upcoming AMD's Llano with 400SPs are practically orders of magnitude more powerful.

Quote from AutoPilot :

I saw that link when Eza first posted it (I've even posted in that thread if you go through it ). I'd say that 4870 is a fair bit faster than ATI 2900, but it's true that LFS is largely CPU limited.
The numbers there are based on the current patch though, unless you think that physics is OK as it is, I still disagree that MT isn't needed. You can always get a bit more refined and more complex model with more CPU performance. And LFS being singlethreaded, there's a LOT more headroom going MT than trying to squeeze more from ST.

The only thread about what I said that I could find in my brief search was this: http://www.lfsforum.net/showthread.php?p=1426006#post1426006

I didn't say it was just DX8, I said it's also outdated, which is true. I mean it ran pretty fine with my Ti4200 back in the day without too many cars around, and IGPs on Sandy Bridge, and especially the upcoming AMD's Llano with 400SPs are practically orders of magnitude more powerful.

Look at my first post in this thread: http://www.lfsforum.net/showthread.php?p=1573560#post1573560. I say that I don't think LFS as-is needs multi-threading. People ask for it without anything else attached, like "I wish LFS had multi-threading and support for better physics and this and that and the other thing". They ask for multi-threading by itself as if it'll make LFS better without tons of other stuff being implemented.

Of course I would love for the physics to improve, actual realistic crash damage, higher poly-count tracks/cars, and tons of other things that would make multi-threading a means to those ends.

Quote from Gener_AL (UK) :

You have tried south city pikd with Full grid?

Well I tried this from the back of a grid at SO4R in GTAL, the lowest my FPS ever goes is 48. This is with all the graphical options in LFS set as high as they can go, and with vsync turned off now. Maybe something is wrong with troy's setup if he's getting as low as 25.

Quote from pik_d :

Look at my first post in this thread: http://www.lfsforum.net/showthread.php?p=1573560#post1573560. I say that I don't think LFS as-is needs multi-threading. People ask for it without anything else attached, like "I wish LFS had multi-threading and support for better physics and this and that and the other thing". They ask for multi-threading by itself as if it'll make LFS better without tons of other stuff being implemented.

Of course I would love for the physics to improve, actual realistic crash damage, higher poly-count tracks/cars, and tons of other things that would make multi-threading a means to those ends.

You said, exact quote, "LFS isn't really at a stage where it needs to use more than one core".

What I had given was a quote from Victor(*) from which it can be inferred that LFS is at a stage where a physics model is too complex to be handled by current CPUs with reasonable FPS, and is one of the reasons for the patch delay. Assuming it's true, do you still think there's no reason for more than one core? I just don't see a rationale for that. Going from one core to four provides a theoretical boost available now that ST performance will not reach in a decade at least.


(*) Shotglass really gave it now that I've read it again, but seems it originates from Victor.

Probably not adding much to this information, But even with my Nvidia Geforce 8400GS/ AMD Anthlon X64 Duo core, And only 2 GB of ram.. I never went below 250ish on FPS with a full grid and the game maxxed, I had it overclocked, Sure, But it proves old technology doesn't suck so bad.

Quote from AutoPilot :

You said, exact quote, "LFS isn't really at a stage where it needs to use more than one core".

What I had given was a quote from Victor(*) from which it can be inferred that LFS is at a stage where a physics model is too complex to be handled by current CPUs with reasonable FPS, and is one of the reasons for the patch delay. Assuming it's true, do you still think there's no reason for more than one core? I just don't see a rationale for that. Going from one core to four provides a theoretical boost available now that ST performance will not reach in a decade at least.


(*) Shotglass really gave it now that I've read it again, but seems it originates from Victor.

I'm talking about what I know, and what people asking for multithreading, Patch Z28. Victor (would be interesting to read that quote if you could paste it here) maybe is talking about what Scawen has behind closed doors? I dunno how you can expect me to be talking about a version of LFS that isn't public if that is indeed what you're talking about with the quote.

Edit: Either way it doesn't much matter, Scawen either will or he wont rewrite stuff to make it multi-threaded, would be nice if LFS got complex enough physics to necessitate MT, I'm sure we can all agree to that (except a handful who have 7 year old computers )

The only reason I brought all this up in the first place is it seems like people are asking for it simply because it's a newer technology, not because LFS actually needs it. Maybe someday it will need it and this discussion will be moot.

I was simply informing you of that possibility, because it was mentioned on this forum, and since I'm hardly a regular here, lots of others must know about it too...

I gave the link to the exact post above.

Oh I see, missed that somehow.