We proceed with a series of practical articles (see also this and this material), in which we analyze sufficiency of various CPU+GPU combinations for games, trying to keep our test conditions as close to popular realia as possible. We hope that such articles will come in handy to readers willing to reasonably distribute their budget between a graphics card and a processor.
This time we've preserved test conditions and consequently the entire configuration from the previous article. By our readers' request we've also added two graphics cards from NVIDIA, the most popular cards in the target price range, and the Intel Pentium E5200 processor. This article is just an update to the review we published in January.
Besides, we added test results of Radeon HD 4830 in the CrossFire mode. The idea of this solution also lies on the surface: a couple of cards with this processor costs as much as a single HD 4870. Besides, you may have to dole out twice as little money at first, which is an attraction for those users, who want to build up performance in future and cannot buy a powerful graphics card right away. Motherboards with two PCI Express x16 slots are available everywhere. They are not much more expensive than usual types, especially those for the AMD platform. Many users buy such motherboards anyway, even if they plan to use just one graphics card, because models with 790GX and 790X chipsets are quipped better and more attractive for a generic home computer than motherboards with cheaper 770 and 780G chipsets. There are some exceptions, of course. For example, we can mention several motherboards with these cheap chipsets supporting CrossFire. But we should keep in mind that the second graphics port in such boards has fewer lanes than the primary slot, which may theoretically reduce performance and lead to minimal FPS drops in some games. What concerns the Intel platform, if you want to assemble a relatively inexpensive computer with CrossFire, you should take a closer look at motherboards with the P45 chipset, those with two graphics ports, of course.
Testbeds
Graphics cards:
Processors:
- Sempron X2 2100 (1.8GHz, 2 x 256 KB L2 Cache, HT1600)
- Athlon X2 4800 (2.5GHz, 2 x 512 KB L2 Cache, HT2000)
- Pentium E5200 (2.5GHz, 2MB L2 Cache, FSB800)
- Athlon X2 6000 (3.0GHz, 2 x 1 MB L2 Cache, HT2000)
- Phenom X3 8750 (2.4GHz, 3 x 512 KB L2 Cache, 2 MB L3 Cache, HT3600)
- Phenom X4 9850 (2.5GHz, 4 x 512 KB L2 Cache, 2 MB L3 Cache, HT4000)
Other parts:
- Motherboards: Gigabyte MA770-DS3H on AMD 770 (all configurations with AMD processors, except HD 4830 CF), Gigabyte MA790X-DS4 on AMD 790X (HD 4830 CF), Gigabyte EP45-DS4 on Intel P45 (all configurations with Pentium E5200)
- RAM: 2 x 1GB DDR2-800 SDRAM Hynix CL5
- HDD: 250GB Seagate Barracuda 7200.10 SATA, 7200 rpm
- PSU: Chieftec 450W
Software:
- Windows Vista Ultimate (32 bit), ATI Catalyst 8.11
- 3DMark Vantage (Default settings, Performance)
- S.T.A.L.K.E.R. 1.003 (GSC Game World/THQ) -- DirectX 9.0, maximum quality settings (dynamic lighting enabled); demo; copy files to the savegames folder, run the game, load level 'ixbt3', and type "demo_play ixbt3" in the console.
- Company Of Heroes: Opposing Fronts (Relic Entertainment/THQ) -- DirectX 9.0, Shaders 2.0, maximum quality settings; run the game, invoke graphics settings and click the Test button.
- Call Of Juarez (Techland/Ubisoft) -- DirectX 10, Shaders 3.0 (HDR); benchmark.
- Crysis (Crytek) -- DirectX 10, High, Crytek Built-in; benchmark
- World in Conflict (Sierra) -- DirectX 10, High, built-in benchmark
- Devil May Cry 4 (Capcom) -- DirectX 10, Maximum, built-in benchmark
- Unreal Tournament 3 (Epic Games) -- DirectX 10, 5 Details Level, 5 Textures Level, CTF-coret; benchmark.
All games were tested at 1680x1050, which is considered the typical resolution for popular 20" LCD monitors. Antialiasing (4x) and anisotropic filtering (16x) options were forced in the drivers.
Tests
We've run the tests for all combinations of graphics cards and processors.
For your convenience, you can use the interactive calculator below:
|
|
 |
|
|
|
