Game tests have been updated right now, because they are more important for chipsets with integrated graphics. But at first we'll publish results in Windows XP and old games, which we used to test motherboards with integrated graphics. These tests would prove that integrated graphics cores were insufficient even for Medium graphics quality settings and smooth gameplay.
| Test |
Intel G35 |
AMD 690G |
GeForce 8200 |
AMD 780G |
| Unreal Tournament 2004 (Medium@800x600), fps |
28.6 |
29.4 |
45.1 |
46.9 |
| Unreal Tournament 2004 (High@1024x768), fps |
25.2 |
27.5 |
35.7 |
38.0 |
| Unreal Tournament 2004 (Highest@1600x1200), fps |
8.7 |
12.0 |
16.3 |
17.7 |
| FarCry (Medium@800x600), fps |
39.6 |
36.2 |
61.0 |
70.0 |
| FarCry (High@1024x768), fps |
33.1 |
30.0 |
41.6 |
48.1 |
| FarCry (Highest@1600x1200), fps |
13.2 |
15.2 |
19.3 |
21.1 |
| DOOM III (Medium@800x600), fps |
13.1 |
14.5 |
40.0 |
43.1 |
| DOOM III (High@1024x768), fps |
12.3 |
12.4 |
26.2 |
29.8 |
| DOOM III (Highest@1600x1200), fps |
5.0 |
5.3 |
12.0 |
13.8 |
These tests illustrate that AMD 780G is more than twice as fast as the previous solution (690G), and it's a tad faster than GeForce 8200. Intel G35 performs approximately on a par with AMD 690G.
Let's proceed to tests in modern games.
| Test |
Intel G35 (Vista) |
NVIDIA GeForce 8200 (Vista) |
AMD 780G (XP) |
AMD 780G (Vista) |
Radeon HD3450 (Vista) |
Hybrid CF (Vista) |
| average fps |
min fps |
average fps |
min fps |
average fps |
min fps |
average fps |
min fps |
average fps |
min fps |
average fps |
min fps |
| 3DMark06 (1280x1024), marks |
619 |
|
1071 |
|
1451 |
|
1547 |
|
1780 |
|
2731 |
|
| Company of Heroes (800x600), fps |
12.5 |
4.0 |
23.5 |
14.0 |
36.6 |
22.0 |
37.6 |
21.7 |
53.9 |
28.8 |
60.5 |
31.9 |
| Company of Heroes (1024x768), fps |
9.7 |
4.0 |
16.8 |
9.7 |
24.5 |
13.7 |
25.6 |
13.6 |
37.2 |
17.7 |
41.3 |
25.2 |
| S.T.A.L.K.E.R. (800x600), fps |
28 |
9 |
60 |
18 |
61 |
15 |
61 |
17 |
87 |
21 |
100 |
19 |
| S.T.A.L.K.E.R. (1024x768), fps |
21 |
8 |
39 |
8 |
40 |
12 |
40 |
14 |
60 |
15 |
68 |
17 |
| F.E.A.R. (800x600), fps |
16 |
7 |
31 |
17 |
29 |
16 |
29 |
17 |
46 |
25 |
52 |
24 |
| F.E.A.R. (1024x768), fps |
11 |
5 |
21 |
10 |
19 |
10 |
19 |
10 |
31 |
16 |
32 |
13 |
| World in Conflict (800x600), fps |
14 |
9 |
32 |
14 |
35 |
14 |
35 |
13 |
53 |
15 |
48 |
18 |
| World in Conflict (1024x768), fps |
11 |
6 |
25 |
10 |
27 |
8 |
26 |
7 |
47 |
13 |
44 |
15 |
| Serious Sam II (800x600), fps |
- |
|
28.6 |
|
31.6 |
|
37.6 |
|
54.2 |
|
61.9 |
|
| Serious Sam II (1024x768), fps |
- |
|
20.7 |
|
24.1 |
|
26.5 |
|
38.3 |
|
45.9 |
|
The layout of forces between the chipsets does not change. As was promised by AMD, you can play even resource-intensive games, even if you have to reduce graphics quality and resolution. As DirectX 9.0c functions are quite sufficient for the graphics quality settings used in our tests, we have tested AMD 780G both in XP and Vista. Performance difference is very small, there are also no differences in image quality - it's a nice surprise, because it hints at well optimized drivers.
Good results with Radeon HD 3450 - we noticed acceleration in all tests. Moreover, minimal frame rates have grown in a number of tests, which we don;t usually expect from CrossFire/SLI solutions.
Hybrid CrossFire does not turn a computer with integrated graphics and a cheap graphics card into a gaming computer. But first of all, it's always nice to be able to pool resources and get performance gains, where it had been previously impossible. Secondly, it's very good performance for the price (a motherboard and a graphics card of this class may be cheaper than a Mid-End graphics card). This solution will be sufficient for modern games. Game developers have been reproaching manufacturers of chipsets with integrated graphics that they don;t even allow to start some games (regardless of their settings) and evaluate how interesting they are. This damages PC positions as a gaming platform. Gamers may decide to buy more serious graphics cards proceeding from their personal preferences rather than from advertisements, no matter how colorful they are. But before they do that, they should have an opportunity to play games even on an inexpensive configuration.
AMD publishes the following 3DMark06 performance results of various graphics cards supporting Hybrid CrossFire.
| Graphics card |
Graphics card alone |
Hybrid CrossFire (AMD 780G + Graphics card) |
| Radeon HD 2400PRO |
1550 |
2600 |
| Radeon HD 2400XT |
2300 |
2700 |
| Radeon HD 3450 |
1768 |
2723 |
| Radeon HD 3470 |
2526 |
2713 |
We can clearly see that the biggest effect is demonstrated when the system pools resources of the chipset and a GPU of similar speed (2400PRO and 3450). The 3450 seems to be an optimal candidate, because this combo demonstrates the maximal total result. It's clear why more powerful cards (3650 and higher) do not support Hybrid CrossFire. Performance gains from this mode will be very small starting with the HD 3470. And if a graphics card scores above 3000 in 3DMark 06, a graphics core in the chipset may even slow down the system because of increased latencies - the chipset core might be too slow to do its share of the load assigned by the driver, while a graphics card is ready to output its share of the work to screen.
|
|
 |
|
|
|
