The results:
F.E.A.R - Performance Test, 1024 x
768
GeForce
8600GTS
GeForce
8600GT
GeForce
7950GT
Radeon
X1950 Pro
F.E.A.R - Performance Test, 1280 x
960
GeForce
8600GTS
GeForce
8600GT
GeForce
7950GT
Radeon
X1950 Pro
F.E.A.R - Performance Test. 1600 x
1200
GeForce
8600GTS
GeForce
8600GT
GeForce
7950GT
Radeon
X1950 Pro
Right off the bat, we can see the GeForce 8600 series does not have what it takes to push this game at our test settings with AA and AF. Anti aliasing performance is definitely not one of the GeForce 8600 series strength. Now, if you look at results without AA and AF, the GeForce 8600GTS is actually slightly faster than the Radeon X1950 Pro at 1024 x 768. At 1280 x 960, none of the cards really have the performance to push frame rates above 30 fps all the time with AA and AF. But then again, all these cards are essentially budget cards. So, if you're planning to buy a GeForce 8600GTS / GT, you must choose between 1024 x 768 with AA and AF or 1280 x 960 / 1024 without AA (at least) and AF. If you chose not to run without AA and AF, even the GeForce 8600GT should be enough. It's slower but the frame rates are playable enough.
Obviously, NVIDIA have to make some compromises to the GeForce 8600 series. That compromise is AA performance - fillrate. Lets examine these scores a little to bit to see how much penalty incurred with AA and AF. First, let's take a look at the cost of angle independent anisotropic filtering. Please note that these results are different from the results used for the graph above. They are from different runs.
8600GTS
| 1xAF (Trilinear) | 2xAF | 4xAF | 8xAF | 16xAF | |
| 1024 | |||||
| Min | 63 | 62 | 62 | 62 | 56 |
| Avg | 119.95 | 120.31 | 120.29 | 117.5 | 114.88 |
| Max | 252 | 251 | 253 | 251 | 249 |
| 1280 | |||||
| Min | 46 | 42 | 43 | 45 | 48 |
| Avg | 86.6 | 85.68 | 84.03 | 82.99 | 83.01 |
| Max | 171 | 174 | 169 | 170 | 169 |
| 1600 | |||||
| Min | 29 | 25 | 27 | 26 | 27 |
| Avg | 58.45 | 58.48 | 58.25 | 56.51 | 56.66 |
| Max | 109 | 109 | 112 | 105 | 104 |
8600GT
| 1xAF (Trilinear) | 2xAF | 4xAF | 8xAF | 16xAF | |
| 1024 | |||||
| Min | 48 | 48 | 45 | 47 | 45 |
| Avg | 94.08 | 92.88 | 92.27 | 92.05 | 90.72 |
| Max | 199 | 196 | 196 | 195 | 192 |
| 1280 | |||||
| Min | 38 | 34 | 32 | 37 | 37 |
| Avg | 66.63 | 66.35 | 65.28 | 65.48 | 65.18 |
| Max | 133 | 133 | 131 | 132 | 132 |
| 1600 | |||||
| Min | 24 | 23 | 24 | 22 | 24 |
| Avg | 45.74 | 45.58 | 44.84 | 44.77 | 44.55 |
| Max | 86 | 86 | 83 | 82 | 82 |
Judging from these test results, anisotropic filtering penalty on the GeForce 8600 is minimal at worst and that's a good thing. Even at high resolutions, the cost is only about 3 to 4 percent drop in frame rates, for both the GeForce 8600GTS and 8600GT. You can actually push anisotropic filtering to the max, with high quality image settings and get some pretty decent frame rates. We can also see, these cards lost quite a bit of performance (25 to 33 percent) when you raise the resolution (both from 1024 x 768 to 1280 x 960 and from 1280 x 960 to 1600 x 1200). It will be interesting to see what kind penalty we can expect with anti aliasing.
8600GTS
| Without AA | 2xAA | 4xAA | |
| 1024 | |||
| Min | 63 | 47 | 29 |
| Avg | 119.54 | 88.3 | 64.44 |
| Max | 255 | 179 | 161 |
| 1280 | |||
| Min | 47 | 30 | 23 |
| Avg | 85.47 | 60.24 | 42.86 |
| Max | 171 | 119 | 100 |
| 1600 | |||
| Min | 29 | 20 | 17 |
| Avg | 57.95 | 40.91 | 28.55 |
| Max | 110 | 79 | 62 |
8600GT
| Without AA | 2xAA | 4xAA | |
| 1024 | |||
| Min | 48 | 35 | 26 |
| Avg | 94.51 | 66.36 | 50.35 |
| Max | 197 | 140 | 122 |
| 1280 | |||
| Min | 38 | 27 | 18 |
| Avg | 68.81 | 46.77 | 33.62 |
| Max | 132 | 93 | 77 |
| 1600 | |||
| Min | 23 | 15 | 12 |
| Avg | 44.4 | 31.25 | 22.21 |
| Max | 86 | 61 | 49 |
Even with the faster GeForce 8600GTS, the penalty of anti aliasing is pretty hefty - frame rates between 2x MSAA at 1024 x 768 and no AA at 1280 x 960 is about the same. This also apply to 4x MSAA at 1024 x 768 and running without AA at 1600 x 1200. Well, running at 1600 x 1200 without AA is actually slightly slower than 4x MSAA at 1024 x 768. Because of the performance penalty, AA is definitely a premium with these cards. Its fairly safe to say that the most optimal solution is to use 2x MSAA (with 16xAF) at 1024 x 768. If you have to use 1280 x 960 / 1024, you'll likely have to forgo using AA to keep frame rates high enough for fluid gameplay. In this respect, there's actually no reason to buy the faster GeForce 8600GTS over the 8600GT.
[Previous Page]
[Go to top]
[Next Page]