Bottleneck?

[DevgruSeal]

Specs listed in my sig, but instead of typing up the technical details, I'll just post a screenshot of CPU-Z readings.

CPU-Z Readings

I'm wondering if I've got a bottleneck here, and what it is.

I seem to not be getting the FPS I expect at certain settings. Things feel quite choppy with STALKER on high-medium settings when I'm not outside, or am in a small village. To be quite honest I expected to at least be able to run it fairly well on high settings. I did expect that framerates would be significantly lower with sun/grass shadows, so I'm not so much worried about that.

I defrag frequently, although I use a free one from Auslogics.

Any thoughts? I don't want to spend mega-bucks on upgrading this machine, because I have a DX10 build planned for sometime around Christmas (Intel x38, nvidia 9 series, maybe the new Intel 45nm transistor CPUs, etc, etc).

 

[Asus]

What resolution and settings do you run it at?

Here are a few reviews that show performance numbers in Stalker, mostly averages.
Graphics cards
CPUs

Shows FPS over time in Stalker. Dips and peaks.

I don't think anything is bottlenecked. The game just demands more of the GPU.
You can see in the CPU benchmark that it didn't benefit going multi-core and not any FPS difference between the CPU speeds. GPU is where it's at for Stalker. Looks like the dynamic light kills FPS vs static lighting. Some pictures between the two on the HardOCP link above.

Didn't Brick upgrade his X1900pro to a 8800 for Stalker? Maybe he can tell you more.

 

[The Brick]

Dynamic lighting is an absolute disaster for your fps, only the 8800s and 2900s can really run it well. But it also scales a lot with the cpu. My e6300 runs at 3.3ghz and it's bottlenecking my 8800gtx in stalker. I indeed upgraded from an 1950xt. It wasn't really worth it, but I was happy with the result.

http://www.digital-daily.com/video/stalker_test

 

[Asus]

Comparing the GPU graphs and CPU graph from the link you gave it looks like it scales with more CPU power at low settings and static light. You might consider that CPU bottlenecked. But it's far from bottlenecked if you run with 1024x768 +dynamic lighting or 1600x1200 +static lighting. The 1024x768 with dynamic light is just CPU scaling. And the 1600x1200 with dynamic light are basically the same and the recorded FPS for each CPU speed are probably all within margin of error.

It's actually just a partial picture of what is happening because they don't try different CPUs with the all the settings used in the GPU graphs. And below with the CPU graph they just use a 8800GTS 320MB. Have to speculate what would happen with different setups.

Found this benchmark which shows all the 8800 versions all really close together at 1280x1024. But the HD 2900XT shows that there is not a CPU bottleneck there since it breaks the mold. If a reviewer just used all 8800 cards you might suspect there was a CPU bottleneck.