S.T.A.L.K.E.R. presents an interesting game test in that it supports 3 rendering modes with rather large differences. The 3 choices of 'Render' modes are: Static Lighting, Object Dynamic Lighting, and Full Dynamic Lighting. Static Lighting is limited to only DX8 with light maps with the in-game options of "Lighting Distance" and "Shadow Quality" disabled. Static Lighting mode gives the game a tremendous performance boost but at a cost. The visuals look flat, washed-out, and dated. Object Dynamic Lighting mode runs in DirectX 9 mode with light maps and selecting Full Dynamic Lighting will run DirectX 9 mode with full dynamic lights. Of course, the last two DX9 lighting modes will place a severe hit on performance compared to the DX8-only supported Static Lighting mode.

In setting up this test, I decided to use the in-game benchmarks provided by the developers. The first, titled 'Buildings Demo' is a timedemo at the military compound where in the game you are given the assignment to free 'Mole.' The second timedemo is in an open area in the first map of the game. Three resolutions of 1024x768, 1280x1024, and 1680x1050 (wide screen) are used with in-game settings as identified below. A minimum of three runs through each benchmark were made with an average recorded as the final result.

S.T.A.L.K.E.R - Settings

S.T.A.L.K.E.R - Buildings Demo Results

S.T.A.L.K.E.R - Short Demo Results

With Static Lighting mode the Inno3D 8600GTS XStriker3 could easily run 100+ FPS with all in-game settings maxed-out at 1680x1050 and can run at higher resolutions. For me the dynamic lighting is a significant part of the game so I did not pursue this any further. Instead, I selected 'Full Dynamic Lighting' mode (this enables 'Lighting Distance' and 'Shadows Quality' in Advanced menu) in the BASIC menu screen and set quality to MAXIMUM (this enables 'Sun Shadow' in Advanced menu). In the ADVANCE menu I set the 'Vision Distance' slider to approximately 60%, 'Objects Detail' to about 85%, 'Grass Density' to a full 100%, 'Textures Detail' to about 75%, 'Anisotropic Filtering' to 100%, dropped 'Antialiasing' slider to '0', activated 'Sun Shadow', de-activated 'Grass Shadow', set 'Lighting Distance' to about 25%, activated 'NPC Flashlights', and set 'Shadows Quality' slider to 100%. At these settings the above results were taken from the two recorded in-game timedemos.

The 'Buildings Demo' with the heavy bump mapping appeared to apply the most stress with the settings that I selected, and of course revealed the lowest FPS results. The 'Short Demo' from the open area timedemo gives an alternative look. A temperature test using nTune in this benchmark revealed an increase in temps from idle of 47C to to a maximum of 54C during the benchmark. At the lowest resolution tested, 1024x768, the 8600GTS XStriker3 breezed through the benchmarks at a pace indicating potential for further tweaking of the in-game settings. Instead, I went on to the 1280x1024 resolution, which I prefer, but found that increasing the 'Vision Distance', the 'Objects Detail', Textures Detail', 'Lighting Distance', and 'Antialiasing' in any combination was just too much for the card to handle, plummeting the frame rates into the 20's.

While the capabilities of the stream processors are becoming more and more apparent, even at a limited number of 32, the performance they generate is outpacing the previous generation and that of the competition. However, the performance the 128-bit memory bus looks to be the hold up when in comes to maintaining maximum settings and upping the resolution in games like S.T.A.L.K.E.R. At 1280x1024 I could find no way to increase 'View Distance' and either 'Objects' or 'Textures Detail' and maintain 'Lighting Distance' and 'Shadows Quality' at present settings. Nor could I increase 'Lighting Distance' above the setting I have selected and no way could I activate 'Grass Shadow' as either just generated too severe a hit at this resolution.

I took a lot of time and went through a lot of trouble to sort this all out to determine what settings I liked best and would work well. If you like the game I think you will agree with me on at least one thing, once you have played the game with 'Full Dynamic Lights' it is difficult to go back to 'Static Lighting' mode (or even DX9 with 'Map Lights') and dropping the resolution is not such a bad idea.

I am impressed with the capabilities of Inno3D's iChill cooling application on the 8600GTS XStriker3. I purposely overclocked the GPU another 9% (761MHz) above the factory setting of 705MHz to generate more heat stress. This tactic did not seem to phase the iChill cooling solution at all, with the maximum recorded temperature of 50C, as you can barely make out, recorded in the bottom nTune graph above rising from the starting idle temp of 44C and tampering off toward the finish of the benchmark. This overclock of the GPU was just for testing the capabilities of the cooling solution on this video card and not used in the benchmarks.

Having logged several gaming sessions with the Inno3D iChill 8600GTS XStriker3, I can say that it has provided enjoyable gameplay at 1280x1024 with the above listed in-game 'tweaked' settings averaging mid-50s FPS while running stable, quiet, and, above all else, staying cool.

Next Page: Conclusion