For our benchmarks, we have chosen the four most popular and widely used game engines. Unreal Tournament 2004, FarCry, Doom 3 and Half-Life 2. Between them, these games and the many games based on the same graphics engines, account, or will soon account for a large proportion of the FPS (First Person Shooter) market.

All these games were tested at three different settings: 1024 x 768 with no AA (antialiasing) or AF (ansiostropic filtering), 1280 x 1024 with x2 AA and x2 AF, and 1600 x 1200 with x4 AA and x4 AF.

For most people, 1280x1024 is the resolution you should be concentrating on, as this is the native resolution of most TFT screens. It is also the best resolution for most CRTs in order to obtain a decent refresh rate.

All games were run with the maximum texture and effects levels, with the exception of Doom 3, which was run with “high” detail settings, since the ultra mode demands 512MB of graphics memory.

Because we have both AGP and PCI-E cards in this test, we wanted our two test rigs to be as close to identical as possible. We used an MSI 865PE Neo 3 with i865 chipset for testing AGP cards, and an MSI 915P Combo with i915P chipset to test PCI-E cards. In both systems we used an Intel Pentium 4 Prescott 3.6GHz. We also used 1GB of CAS 2.5 Kingmax PC3200 running in Dual Channel mode.

We know this system is probably beyond the specification of what most of you can afford, but we wanted the CPU to hold back this cards as little as possible. If running an older system, expect as much as 30% performance decrease.

We used nVidia Forceware 66.93 drivers and ATI Catalyst 5.1 drivers. Everything else was kept the same.

Take note of the “value for money” rating given to each card. This is the amount of 3D Marks attained per pound, which gives a comparative measurement of value for money.

<< Back | 1 | | 2 | | 3 | | 4 | | 5 | | 6 | | 7 | | 8 | | 9 | | 10 | | 11 | | 12 | | 13 | | 14 | | 15 |
| 16 | | 17 | Next >>