Games Benchmarks - OpenGL
Doom 3 "Turkey Baster"
Our Doom 3 "Turkey Baster" test uses an extended sequence from the Alpha
Labs 1 level and is intended to be used in Doom 3's Ultra mode as a slightly
heavier stress test as this level has a few areas where texture swapping will
occur on 256MB boards.
Doom3 "Turkey Baster" (FPS) |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XT |
140.2 |
141.6 |
140.7 |
121.2 |
95.9 |
X1800 XL |
137.4 |
135.5 |
120.5 |
90.4 |
68.7 |
X850 XT PE |
139.2 |
130.2 |
122.5 |
92.1 |
70.2 |
X800 XT |
139.2 |
132.7 |
110.5 |
82.8 |
62.8 |
X1800 XT % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XL |
2.0% |
4.5% |
16.8% |
34.1% |
39.6% |
X850 XT PE |
0.7% |
8.8% |
14.9% |
31.6% |
36.6% |
X800 XT |
0.7% |
6.7% |
27.3% |
46.4% |
52.7% |
X1800 XL % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X850 XT PE |
-1.3% |
4.1% |
-1.6% |
-1.8% |
-2.1% |
X800 XT |
-1.3% |
2.1% |
9.0% |
9.2% |
9.4% |
Under this Doom 3 test the extra 512MB of RAM doesn't appear to be making too
much of a difference as the performance gain the X1800 XT has over the X850 XT
PE is in a similar range to other titles that we've seen so far. In this
instance the X1800 XL's performance is up to 9% higher than the X800 XT, showing
that the improvements made to the architecture have carried across to Doom 3's
utilisation patterns, but not necessarily to the same degree as the X850 XT PE
is slightly faster.
Doom3 "Turkey Baster", 4x FSAA & 8x AF (FPS) |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XT |
137.7 |
124.8 |
101.8 |
72.2 |
52.7 |
X1800 XL |
120.9 |
99.6 |
79.0 |
55.8 |
41.7 |
X850 XT PE |
124.6 |
104.9 |
85.2 |
61.7 |
46.1 |
X800 XT |
114.7 |
94.3 |
75.8 |
54.6 |
41.1 |
X1800 XT % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XL |
13.9% |
25.3% |
28.9% |
29.4% |
26.4% |
X850 XT PE |
10.5% |
19.0% |
19.5% |
17.0% |
14.3% |
X800 XT |
20.1% |
32.3% |
34.3% |
32.2% |
28.2% |
X1800 XL % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X850 XT PE |
-3.0% |
-5.1% |
-7.3% |
-9.6% |
-9.5% |
X800 XT |
5.4% |
5.6% |
4.2% |
2.2% |
1.5% |
With 4x FSAA and 8x AF enabled the X1800 XT is, as we'd expect, still
fastest, but curiously not to the same degree as we've seen in other cases with
these settings. This is underlined by the case that the X1800 XL has a very slim
performance margin over the X800 XT.
Doom3 "Turkey Baster", 6x FSAA & 16x AF (FPS) |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XT |
136.5 |
122.8 |
100.8 |
72.5 |
53.0 |
X1800 XL |
112.5 |
92.8 |
73.6 |
52.6 |
34.6 |
X850 XT PE |
98.1 |
77.9 |
60.2 |
42.3 |
28.0 |
X800 XT |
88.4 |
70.0 |
53.7 |
37.9 |
25.4 |
X1800 XT % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X1800 XL |
21.3% |
32.3% |
37.0% |
37.8% |
53.2% |
X850 XT PE |
39.1% |
57.6% |
67.4% |
71.4% |
89.3% |
X800 XT |
54.4% |
75.4% |
87.7% |
91.3% |
108.7% |
X1800 XL % Faster than: |
640x480 |
800x600 |
1024x768 |
1280x1024 |
1600x1200 |
X850 XT PE |
14.7% |
19.1% |
22.3% |
24.3% |
23.6% |
X800 XT |
27.3% |
32.6% |
37.1% |
38.8% |
36.2% |
With 6x FSAA and 8x AF enabled the performance differentials are about where
we would expect them, given the previous patterns. In this case we see that the
X1800 XL has as much as a 39% performance improvement over the X800 XT.
Curiously, the result at 1600x1200 for the X1800 XT here is actually faster than
at 4x FSAA & 8x AF, despite more FSAA and texture samples being calculated in
this case.