How we test
GPU comparisons |
|||||||
|---|---|---|---|---|---|---|---|
| Graphics card | Approx. pricing |
GPU clock (MHz) |
Stream processors |
Shader clock (MHz) |
Memory clock (MHz) |
Memory bus (bits) |
Graphics driver |
| Gigabyte GeForce GTX 590 3,072MB and 'OC' | £605 | 675 | 1,024 | 1,350 | 3,700 | 384 x 2 | ForceWare 267.84 |
| Gigabyte GeForce GTX 590 3,072MB | £605 | 607 | 1,024 | 1,214 | 3,414 | 384 x 2 | ForceWare 267.84 |
| ASUS GeForce GTX 580 SLI | £770 | 772 | 512 | 1,544 | 4,008 | 384 x 2 | ForceWare 266.58 |
| ASUS GeForce GTX 570 SLI | £540 | 732 | 960 | 1,464 | 3,600 | 320 x 2 | ForceWare 266.58 |
| ASUS GeForce GTX 580 1,536MB | £385 | 772 | 1,024 | 1,544 | 4,008 | 384 | ForceWare 266.58 |
| ASUS GeForce GTX 570 1,280MB | £270 | 732 | 480 | 1,464 | 3,800 | 320 | ForceWare 266.58 |
| Sapphire Radeon HD 6990 4,096MB OC | £540 | 880 | 3,072 | 880 | 5,000 | 256 x 2 | Catalyst 11.4 |
| Sapphire Radeon HD 6990 4,096MB | £540 | 830 | 3,072 | 830 | 5,000 | 256 x 2 | Catalyst 11.4 |
| HIS Radeon HD 6970 XF | £570 | 880 | 3,072 | 880 | 5,500 | 256 x 2 | Catalyst 11.4 | Sapphire Radeon HD 6950 XF | £410 | 800 | 2,816 | 800 | 5,000 | 256 x 2 | Catalyst 11.4 | HIS Radeon HD 6970 2,048 | £285 | 880 | 1,536 | 880 | 5,500 | 256 | Catalyst 11.4 | Sapphire Radeon HD 6950 2,048 | £205 | 800 | 1,408 | 800 | 5,000 | 256 | Catalyst 11.4 | Sapphire Radeon HD 5970 2,048 | £500 | 725 | 3,200 | 725 | 4,000 | 256 x 2 | Catalyst 11.4 |
Test bench |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CPU | Intel Core i7 980X (3.33GHz, 12MB L3 cache, six-core, LGA1366) | |||||||||||
| Motherboard | ASUS P6X58D-E (0502 BIOS) | |||||||||||
| Memory | 6GB Corsair Dominator (9-9-9-24-2T @ 1,600MHz) | |||||||||||
| Power Supply | Corsair HX1000W | |||||||||||
| Monitor | Dell 30in 3007WFP | |||||||||||
| Disk drive(s) | Crucial RealSSD C300 (256GB) | |||||||||||
| Chassis | Corsair Obsidian Series 700D | |||||||||||
| Operating system | Windows 7, SP1, 64-bit | |||||||||||
Benchmarks |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3DMark 11 | Extreme preset, overall score and combined test results recorded. | |||||||||||
| Aliens vs. Predator | DX11, 1,920x1,080 and 2,560x1,600 resolutions, 4xAA, 16xAF, very high quality. | |||||||||||
| Call of Duty: Black Ops | DX9, 1,920x1,080 and 2,560x1,600 resolutions, 8xAA, ultra quality, FRAPS-recorded benchmark. | |||||||||||
| F1 2010 | DX11, 1,920x1,080 and 2,560x1,600 resolutions, 8xAA, ultra quality, Monaco circuit. | |||||||||||
| Just Cause 2 | DX10, 1,920x1,080 and 2,560x1,600 resolutions, 4xAA, 16xAF, very high quality, Dark Tower benchmark. | |||||||||||
| Metro 2033 | DX11, 1,920x1,080 and 2,560x1,600 resolutions, 4xAA, 16xAF, very high quality. | |||||||||||
| StarCraft II | DX9, 1,920x1,080 and 2,560x1,600 resolutions, 4xAA, ultra quality. | |||||||||||
| Power consumption | To emulate real-world usage scenarios, we record mains power draw both when idle and whilst playing Just Cause 2. | |||||||||||
| Temperature | To emulate real-world usage scenarios, we record GPU core temperature both when idle and whilst playing Just Cause 2. | |||||||||||
| Noise | A PCE-318 noise level meter is placed at front of a Corsair 700D chassis with side panel on. | |||||||||||
Notes
We feel there's little point in benchmarking the card at the default speed alone; we've done that in the original review. Since our initial examination last week, NVIDIA has launched the ForceWare 267.84 drivers. Benchmarking the Gigabyte on the press-supplied and publicly-available sets shows no difference in performance across our seven benchmarks.
Instead, to keep it interesting for you and provide a glimpse of where the GeForce GTX 590 is heading once partners have had time to tinker with it, we overclocked the Gigabyte card to what can be considered a moderate level.
Now, there's a huge disclaimer coming up. Ready for it?
We do not guarantee that any GeForce GTX 590 card will be able to operate at these frequencies on a long-term basis. Overclocking may be detrimental to your GTX 590's health, and any such overclocking is undertaken at the sole risk of the user.
Phewee. We managed to hit the above-shown 675MHz core and 3,700MHz memory frequencies without meddling with the voltage, and the card behaved impeccably throughout testing.
The purpose of this overclock is to show the GTX 590's true potential, rather than the conservative clocks programmed in by NVIDIA.
You'll see two GTX 590s in the upcoming graphs. The dark-green bar relates to the Gigabyte's stock-clocked performance, while the light-green bar shows what happens when clocks are dialled up.
Note, too, the HEXUS-titled Gigabyte GTX 590 'OC' is not a production card; you cannot buy it. It's worth repeating that the sole purpose is to show how the GTX 590's performance scales with a moderate overclock in place.
