In figuring out what I wanted to do with the tests for this review, I decided to choose games and benchmarks that show performance differences with SLI. After all, if SLI isn’t supported, there’s not much sense in running in SLI mode. The second consideration is that many older applications are CPU limited even with one 6800GT. The last consideration is that running benchmarks at 1024x768 is a limitation I decided to work around for this review. To reflect the increased performance of two cards, I decided to test the benchmarks at 1280x1024 resolution without FSAA and Anisotropic Filtering and with 4x FSAA and 8X AF. If you’re buying a 6800GT SLI system, you’re not likely to want to play games at 1024x768. Unfortunately, my 19” LCD monitor has a maximum resolution of only 1280x1024; otherwise I’d do some 1600x1200 benchmarks as well.

The test system is an interesting mix of the latest available hardware. I don’t have a FX-55 CPU to test, so a 4000+, which we use in our 939 platform was used. This CPU is the 2nd fastest available AMD CPU on the market. It has 1 megabyte of L2 cache, and runs at 2.4 GHz, which is equal to the 3800+. The Gigabyte GA-K8NXP SLI board provided worked like a charm.

Test System
AMD Athlon 64 4000+ running 2.4 GHz
1 GB Crucial Tracer memory running 400 MHz CAS 2.0
6800GT running 66.93 Forceware drivers
Dual 6800GT running 66.93 Forceware drivers
80 GB Seagate Barracuda IV 8MB cache SATA HDD
Gigabyte GA-K8N SLI motherboard
Windows XP with SP2
DirectX 9.0c

Test Software
Halo
Half Life 2
Doom 3
3dmark05
3dmark03