Had the problem, I have Radeon x850 Pro, PCI-E. With 5.13 LFS was really slow, 25-30 fps and in starts lower than 15. Works well with 5.11 now.

EDIT:
And my comp:
3200+ (2,2ghz) Athlon
1536 mb RAM
x850

At first I wasn't going to update, but in the name of LFS I'll test this out tomorrow after golf.

CPU: Athlon XP @ 2350mhz
GFX: 9800pro

Edit: I'm no longer going to test this out. Not only do I not have HT, the compulsary CCC inclusion in the new driver packages is bound to screw up something on my computer. Windows Update has issues installing .NET service packs on my computer, and as such I wouldn't be in the best position to give reliable results about the latest drivers.

Please anyone got a link to a working 5.11 or bellow as i cant play south city trying turn one at 10 - 15 fps is awful

9500pro athlon xp 2800+ seems to work fine (omega drivers)

Quote from Billeh :

Please anyone got a link to a working 5.11 or bellow as i cant play south city trying turn one at 10 - 15 fps is awful

http://downloads.guru3d.com/download.php?id=18

der_jackal and Qurpiz : you are the only ones so far with non-Intel CPU who are reporting the same problem, so that's interesting.

der_jackal : AMD64x2 4800+
Qurpiz : 3200+ (2,2ghz) Athlon

Do these processors have a similar feature to Hyper-Threading - perhaps under a different name? AMD64x2 : is this actually two separate processors, or a single HT processor?

the x2 series are two completely independant cores, appears and functions the same as a dual cpu system.

the 3200 wont be a x2, also amd dont have an HT equivalent so the 3200 will be a single core (3800x2 is the first in the x2 series)

also i think the 3200 is 2.0ghz as my 3000 is 1.8 @2.55Ghz (maybe its an older model cpu or not amd64)

Troy

Quote from Scawen :

der_jackal and Qurpiz : you are the only ones so far with non-Intel CPU who are reporting the same problem, so that's interesting.

der_jackal : AMD64x2 4800+
Qurpiz : 3200+ (2,2ghz) Athlon

Do these processors have a similar feature to Hyper-Threading - perhaps under a different name? AMD64x2 : is this actually two separate processors, or a single HT processor?

From AMD's FAQ section:

Q:Does the AMD Athlon 64 X2 Dual-Core processor design have two AMD Athlon 64 processors on a single die?
A:Yes. The AMD Athlon 64 X2 Dual-Core processor design is true dual-core - with two CPU cores on one die. AMD’s dual-core architecture relies on advanced Direct Connect Architecture to supply each core with the memory and I/O bandwidth needed to maintain high performance per core. AMD’s on-chip integrated memory controller ensures that both cores have optimal access to the integrated memory controller and integrated HyperTransport™ links, so that performance scales well with the number of cores.

They also have something called HyperTransport technology which I don't think is the same as HT for intel they probably just named it similar for maketing purposes...

Having said that the current generation of P4's which are the P4D range are also true dual core processors, a quote from there sales guff "Intel® Pentium® D processor is Intel’s preferred desktop
processor. Because it is powered by two execution cores in one processor"

I'm curently running with an P4D 840 3.2 but unfortunately have an nvidia card so can't see if it affects the ati cards the same as the old P4 technology... is it possible it's not about HT at all but the way dual core processors (virtual or not) organise thread processes?

It is also apparent that Intel is using the same or at least a modified version of there HT technology on the new dual core processors they offer...

HyperTransport is just the connection bus (its an open standard - not just amd, but ofcause intel wont sign up to standards as usual)

Troy

Whilst Hyper-Threading is a psuedo dual core system, I think the OS (and more than likely the GFX drivers) see it as 2 physical CPU's.
Might be why there is that link between the X2 having the same issue.

Dan,

Additional info is that 5.12/5.13 drivers are 'multicore' optimized

The psuedo nature of the HT is that it only has partial genuine simultainious processing capability (only part of the hardware is duplicated) and can run anywhere from 45% to 75% of a genuine dual system of equal type.

Troy

I think it's pretty obvious now that HT/DualCore is the source of all these problems. I doubt ATi did anything wrong in their drivers, I rather tend to believe that it is LFS which is messing things up. This first showed when mighty powerful HT/DualCore CPUs where generally a bit slower than equally or lower rated single core CPUs, although the effect was not this strong (IIRC).

Now, the only major changes in the new .12/.13 ATi drivers is optimized (or added?) HT/DualCore handling. It seems that the little handicap HT/DC CPUs had prior now get really noticable because the GFX card tries to mess around with the other core too, causing some quirk with LFS somewhere deep in the drivers.

Maybe "just" making LFS use the other core too (for some misc calculations) would solve these problems?

Opteron 1.8Ghz S939 Single Core
Sapphire X800GTO²
5.13

excellent performance with every detail maxxed out

As I stated earlier in this thread I currently find that LFS runs better on my system (stable) with the lfs.exe afinity set to cpu1 and its priority set to system process (highest) so would be interesting if anyone with an ati card can see if this helps the situation at all

Information that may be relivent or not (to do with threading)

Im a A64, single core and an nvidia card

when a background process comes 'alive' such as my norton updating, in say fear or hl2 i get a little 'stuttering' and a performance fps drop of maybe 25%, if the same happens in LFS i get a performance drop of 90% (im locked at 100 fps normally and get <10 fps when 'something' happens in the background) maybe its just a LFS thread priority issue?

Im sorry if this post is not usefull or taking the focus away from the real problem, just thought it may be relivent.

Troy

Quote from AndroidXP :

I think it's pretty obvious now that HT/DualCore is the source of all these problems. I doubt ATi did anything wrong in their drivers, I rather tend to believe that it is LFS which is messing things up....

But then why would it all be working perfectly well, above 100 fps with the 5.11 drivers, and drop so terribly to 10 fps with the 5.12 / 5.13 drivers?

Why should that be LFS - when the only difference is the drivers?

Of course, LFS is involved somewhere here, because this problem isn't affecting some other games, those that were tested by ATI, for example. But i don't see anything here that indicates that LFS is doing anything wrong, other than perhaps using something in DirectX 8.1 that ATI didn't include in their test program?

One more question :

For those who experience the massive drop in frame rate, i guess the frame rate is still very low, even if you press the pause key? Is that true?

I'm asking that just to be sure that the updated drivers aren't somehow causing LFS physics to run slowly, and so the slowdown is almost certainly in the drivers themselves. If that's quite well established then the ATI guys will really need to run LFS over a debug version of their drivers on a HT system and they may be able to see instantly where it's tripping up.

Sometimes there is a huge slowdown due to exception handling. The scenario for that could be that somehow there is an error (or many errors) each frame due to a certain DirectX function called by LFS, and these could be causing the slowdown.

Scawen, i would agree with your general statement, because nvidia's latest drivers are dual core optimized (but not for ht) and nobody seems to be reporting the same issues with LFS

Troy

More info :



I was reading about the ati drivers when they came out, i know there is a registry setting to disable the dualcore/ht support in the drivers but i cant find it again

Im sure it was on www.rage3d.com forums

Troy

I understand you and I really have no idea how (if?) code has to be modified to be HT/DualCore "compatible". It's just that I have yet to hear of any game besides LFS that actually suffered from the new drivers.

I wanted to say that I doubt LFS has problems with the ATi drivers, but LFS has problems with HT/DC. The new ATi drivers trying to use HT/DC exaggerates this problem big time.

The first logical thing is of course to blame ATi for the problems because they first arised when their drivers changed (and to some extent it's indeed their drivers "causing" the problem). But what should not be overseen is that they don't "cause" the problem, they just make it more apparent. Now you can argue that the "root problem" was small enough to be ignored and not considered a problem, but ideally we should find out why LFS has these problems with HT/DC and solve them. Most probably the .12/.13 problems will be solved aswell.

E: heh yeah. try{ } catch { } is awfully slow And I really have no idea how your code looks so I think wild guesswork won't get us anywhere. Either you find something "by luck" and it solves the problem or ATi can be arsed to actually debug their drivers running LFS which would be way more helpful

Is there no debugging system we (those who have problems) can run whilst LFS is running to try and pin point the problem?.
I am 99% sure that its a HT/Dual core issue, as I have seen it rear its ugly head many times.

Dan,

Hmm, i just stumbled over this

Quote :

Detecting HT Technology or dual cores requires the use of the CPUID instruction. [...] The CPUID instruction is not visible to high-level languages such as C/C++, C#, or Java. Calling the CPUID from these languages requires either the support for inline assembly or the ability to call an assembly language routine.

I guess you can't really do that much code fixing regarding HT/DC by yourself. It really seems to be a by-product of some weird driver combinations and/or bad luck.

Just experienced something new,

when i run lfs in windowed mode, everything is fine - until i (for example) press the start menu, or or bring up any other "right-click" menus ... Then the fps drops 'til <10. the fps stays low even when i pause the game. It doesn't seem to have to do with lfs losing focus, i can use other applications and the game will run fine, until i use those menus, or a "hint" pops up (if i place the cursor of the clock for example, a "hint" will show me what date it is.)

det 5.13, x800 @ win xp x64 ...

EDIT: woops, found a solution - disabling "shadows under menus" in apperance/effects fixed the problem the menus, the "hints" still slow things down though. please let me know if you know a way to correct this ..

ATI X850 XT PE
Athlon Barton 2800+

No noticeable difference in fps.

For some reason the AF doesn't work anymore, though illepall And on windows startup I get an error message "cli.exe" couldn't be initialized. Hopefully switching back to 5.11 will solve this

P-IV 3GHz HT, 1024 DDR-400, ATi Radeon 9800 Pro 128MB

With the new "Omega 2.6.87" couldnt get over 18fps,
where I had 90fps with the previous Omega 2.6.71 before.

Tried everthing for a while, then decided to step back to previous version of drivers.