nKP 1.1 and ATi graphics card preformance

[Guest lovretta]

For the record, i just did test beta2 and i get around 100 fps on spa.

[Guest lovretta]

More findings:

1st

[Guest lovretta]

Right now i was online with F1600 @ Prato Short, grid of 10 drivers. One of them mentioned "10 fps". I Asked "Ati?" he said "yes". Dunno who it is, his nickname was something like "A-GT" - if you read this, write your pc specs and rest of the stuff here.

[Guest NightShift]

AMD Athlon 64 X2 Dual Core 4200+ 2.21 GHz, 2GB RAM, ATI Radeon HD 4670 512MB (10.3 driver), Windows XP 32bit

My simracing PC is similar to this machine, so out of curiosity I decided to do some tests, trying to replicate Matt's system settings as far as possible (some he didn't post so I had to fill them in myself). I've found something that may be interesting. Or not. You tell me.

500 AC, Spa, outside the pits, minimum FPS:

CPU Speed            FPS          CPU Delta       FPS Delta
2100 MHz             11           --              --
2400 MHz             18           +14%            +63%
2700 MHz             32           +28%            +190%

Weird ain't it.

Edited April 19, 2010 by NightShift

[Guest lovretta]

So if i got it right, you overclocked your CPU and got higher numbers?

I tried to do so myself, but due to my crappy mainboard, i can't do much with my cpu. Default FSB is 200, got it up to 215 MHz, finally got 3,24 GHz CPU speed.

Just after that, i tried spa, and got a 17 fps in pits. Thats a really high increase, as before i had 5 fps in pits.

Also tried with Hyper Threading disabled and fps dropped to 9 (overclocked cpu that is).

And i overclocked my VGA, default clocks 575/700 to 600 / 800. Really marginal increase on core clock, so as expected didn't help by any means.

Whish i had a better mainboard, i'm really courious wheter overclocking further would help.

I got an idea out of all this - .net framework while installing "optimizes" runtime for your pc - could it be due to this?

[Guest NightShift]

So if i got it right, you overclocked your CPU and got higher numbers?

Actually the CPU was underclocked to try and match the performance of Matt's computer.

What I found interesting is: 1- the non-linear relationship between frequency and FPS and 2- ...

I got an idea out of all this - .net framework while installing "optimizes" runtime for your pc - could it be due to this?

...I did complete lap around Spa with the CPU at full speed and the FPS minimum is on the other side of the track: so I would expect the FPS on my PC to go pretty close to the 5 FPS minimum that Matt has recorded on his.

So I think there's a sort of threshold for absolute computing power under which the PC struggles, and above that the situation improves very quickly.

[Guest Katsushika Matsumoto]

I've found something that may be interesting. Or not. You tell me.

Weird ain't it.

Yeah that's weird.

I did monitor my CPU usage when running nKP 1.1 and it's running 80% (both cores) at max.

(E7300 @ stock)

Edited April 20, 2010 by Katsushika Matsumoto

[Guest NightShift]

BTW I'm also using an ATI card with the latest catalyst and I'm not quite satisfied with the quality of the output

I think the quality problems I've been experiencing go hand in hand with the FPS capping bug. I've been paying closer attention to the FPS, so I've noticed the capping can happen at relatively high rates: it's not just a matter of 16 or 32, it can even get capped at 63.

Whenever that happens it seems filtering is not applied as it should. A reboot fixes both the FPS problem and the issue with quality so if in doubt, it's something worth trying.

The problem however lies with netKar it's annoying enough have to close down all of my apps just because nKP is not behaving.

Edited April 23, 2010 by NightShift

[Julian B.]

The problem however lies with netKar it's annoying enough have to close down all of my apps just because nKP is not behaving.

This helps a little bit but not enough to drive normal like in the beta.

[Guest Matt Dawson]

i suspect it might be due a new and more refined wheel/road intersection routine.. it didn't cause any problems in any of our test machines. i have a version out to test right now to prove if the problem is actually there.

Any news on this?