Game Unplayable

[KG26_Alpha]

Nice first post.
Welcome to the forums.

[Rattlehead]

RE77Action - why are you running model detail on only medium? You should be able to put it on high with little to no ill effect.

[RE77ACTION]

I've model detail on medium because it gives me one FPS extra in Black Death (average 32 vs 30/31). When testing the FPS in black death I use every time the following consistent procedure. Every test I start the game clean from Windows, load and start the track for 5 seconds and then exit the track. Then load it again, wait 5 seconds and start the track for the second time. From this run I take the average from the absolute beginning (includes building the screen and loading all textures) to the absolute end. The reason for this is because I've learned that the game is not always consistent in its FPS returns when running each benchmark right after the other. Another reason is that I like to restart the game when changing settings, even when the game itself doesn't ask for it. This is just to be sure.

The biggest performance hit I get from shadows. Shadows cost me easily 30-40% in frame rate. However, I think it's worth it because it adds enormous to the overall experience.

I've played extensively with all the settings and I have to get my FPS from somewhere. It's a tradeoff that I'm willing to make because the loss in quality between medium and high is in my opinion minor. But I clearly understand that others would make other choices on personal preference and hardware.

[RE77ACTION]

@TonyD:

Thanks for your experience with upping your CPU performance! My CPU isn't the strongest one when it comes to OC. With 3,42 Ghz @ 1,525V (stable) I'm at my absolute max unfortunately. I which I could go higher to test the difference. Typing this gives me new inspiration, because I could of course go lower to test the difference. I'll be back later today to give my results.

[RE77ACTION]

As promised I've tested the Black Death track several times with scaled down CPU performance. For this test I've lowered the multiplier of my CPU from 15 to 13 and kept everything else the same (FSB still 227+1 (because of smiley error when I type eight in numbers). This way we know we are only testing the CPU and not things like memory, NB, SB, etc. The CPU runs now with 4 cores on 2,96 GHz instead my originally overclocked 3,42 GHz.

The results are pretty staggering if you ask me. The average frame rate dropped from 32 to 29 which is a decrease of about 9.5% while I decreased the CPU speed with about 13,5%. The strange thing here is that the CPU utilization is about the same as it was with higher CPU speeds. And still none of the cores reached ever 100% for a moment.

Maybe I and many others get never 100% utilization shown because the game uses functions which saturates only a part of the processor. And saturating only that part never translates to full utilization in a graph. I'm no expert on this, but I vaguely remind that I once read something like this.

It looks to me that my processor is a bottleneck even when it never shows a full 100% utilization in any one of its cores. In these tests I've grown stronger towards a better CPU. However more CPU power will give diminishing returns as this little test already showed.

I hope someone else will find these results useful.

[TonyD]

Originally Posted by RE77ACTION

...
The average frame rate dropped from 32 to 29 which is a decrease of about 9.5% while I decreased the CPU speed with about 13,5%. ...

I was a bit surprised to see the difference mine produced at a higher speed, the feeling in the game is not nearly as apparent as the result would suggest. Although a 7.3% increase in average frame rate is not huge, especially for a ~17% increase in cpu speed, it could be beneficial if frame rates were marginal to start with. Another benefit of the cpu’s that you are considering is the fact that they have unlocked multipliers, which is handy when overclocking. I don’t really use this feature since it negates using Cool’n’Quiet – I tend to leave it standard (my RAM won’t run much faster than 533MHz, which it does by default on my mobo, and is faster this way than running at 400MHz and then increasing the ‘FSB’).

An upgrade I intend doing shortly is replacing my gfx card with a 6950 or 6970, which is due, but am going to wait for Bulldozer before deciding on a new cpu/mobo/RAM (currently leaning towards SandyBridge, although I'm an AMD fan). I just hope that some of issues that the game has with these cards are resolved by then, as my current one works flawlessly.

Good luck with your decision

[RE77ACTION]

@TonyD and others:

That's exactly what I meant with diminishing returns. Your CPU (although the same operating clock speed) is better because its more efficient while your graphics card is a little less (no offence). It's to be expected that you will get less FPS increase when raising your CPU speed. I expect the difference will get bigger when you upgrade to a 6950 or 6970.

After seeing TUSA/TX-Gunslingers beautiful post in another tread (http://forum.1cpublishing.eu/showthread.php?t=22417) I saw that his GPU (5870) peaks often at 100% for longer periods while mine peaks at about 71% for only very short periods. If I can lift this percentage to 100%, this would mean at least a 40% increase in performance which would be clearly noticeable in a heavy game like this. Besides, I can use the power for other things too (like heavy multilayer Photoshopping). So, I've enough excuses . However I will probably first await the next patch.