System setup and notes
Hardware
| Graphics cards | GeCube Gemini X1650 XT 512MiB | ASUS Radeon EAX1950PRO 256MiB | ASUS EN8600GTS 256MiB | Inno3D iChiLL 7900GS Arctic Cooling Silencer 6 |
|---|---|---|---|---|
| Shader Model | 3.0 | 3.0 | 4.0 | 3.0 |
| GPU Clock Speed (MHz) | 574 | 580.5 | 675 | 550 |
| Shader Clock Speed (MHz) | 574 | 580.5 | 1450 | 550 |
| Memory Clock Speed (MHz) | 1377 | 1404 | 2016 | 1500 |
| Memory Bus Width (Bits) | 128 | 256 | 128 | 256 |
| CPU | Intel Core 2 Duo E6600 (2.40GHz, 4MiB L2 cache, 1066MHz FSB, LGA775) | |||
| Motherboard | ASUS PW5-DH Deluxe (Intel i975X) | EVGA NF68 (NVIDIA nForce 680i SLI) | ||
| BIOS revision | 1305 | 691N0P20 | ||
| Memory | 1GByte (2 x 512MByte) OCZ26671024ELDCGE-K PC5400 | |||
| Memory timings and speed | 4-4-4-8 @ DDR2-667 | |||
| Disk drive(s) | 160GB Seagate Barracuda 7200.9 (3Gb/s mode) | |||
| Mainboard software | Intel Inf Update 8.0.1.1002 | NVIDIA platform driver 9.53 | ||
| Graphics driver | CATALYST 7.1 | ForceWare 158.22 | ForceWare 93.71 | |
| Operating System | Windows XP Professional, w/ SP2, 32-bit | |||
Software
| 3D Benchmarks |
Far Cry v1.33 Quake 4 v1.30 Splinter Cell: Chaos Theory v1.05 |
|---|
Notes
Representing two Radeon X1650 XTs and being priced at around £120, we're comparing the GeCube X1650 Dual against a couple of established ~£100 SM3.0 GPUs that have been our favourites for a while. An ASUS GeForce 8600 GTS, run off newer technology, is also added in to see how DX10 cards cope with our older titles.It's a pre-CrossFired setup against DX9 GPUs derived from, at the time, class-leading architectures.
HEXUS took a look at the performance of a couple of ASUS Radeon X1650 XTs in CrossFire here. The single-card GeCube X1650 XT Dual 512MiB offers near-identical performance, as you'd expect.
Benchmarks were conducted at 1280x1024 4xAA 8xAF and 1600x1200 0xAA 8xAF. These are the most likely settings for resolution and image-quality on mid-range cards, especially given that they're the native resolutions of many reasonably-priced LCD monitors.
As always, we ran each benchmark a trio of times and then calculated the arithmetic mean. If any of the three results looked erroneous, we threw all three away until we could collect three within a margin of statistical error. We report any major attempts needed to get three reliable results, of course. Apart from that, things are as noted on the graphs and in the graph commentary. Want to know more? Head for the HEXUS.community.
