To ensure that all reviews on Overclock3D are fair, consistent and unbiased, a standard set of hardware and software is used whenever possible during the comparative testing of two or more products. The configurations used in this review can be seen below:
CPU: Intel Nehalem i7 920 Skt1366 2.66GHz (@3.8 Ghz)
Motherboard: Gigabyte EX58-UD5
Memory: 3x2GB Corsair DDR3 1600mhz @ 8-8-8-24
HD : Hitachi Deskstar 7k160 7200rpm 80GB
GPU: Asus ENGTX275
Graphics Drivers: Supplied by Asus
PSU: Gigabyte ODIN 1200w
During the testing of the setups above, special care was taken to ensure that the BIOS settings used matched whenever possible. A fresh install of Windows Vista was also used before the benchmarking began, with a full defrag of the hard drive once all the drivers and software were installed, preventing any possible performance issues due to leftover drivers from the previous motherboard installations. For the 3DMark and gaming tests a single card configuration was used.
To guarantee a broad range of results, the following benchmark utilities were used:
3D / Rendering Benchmarks
• 3DMark 05
• 3DMark 06
• 3DMark Vantage
• Far Cry 2
• Race drive: GRID
• Call of Duty IV
• Unreal Tournament III
Power consumption was measured at the socket using a plug-in mains power and energy monitor. Because of this, the readings below are of the total system, not just the GPU. Idle readings were taken after 5 minutes in Windows. Load readings were taken during a run of Furmark.
The power consumption of the GTX275 was nothing out of the ordinary falling into the middle of the pack on both idle and load configurations. As with most of the GPU's on test, the more powerfull the card, the greater the amount of consumption.
Temperatures were taken at the factory clocked speed during idle in Windows and after 10 minutes of running Furmark with settings maxed out (2560x1600 8xMSAA). Ambient temperatures were taken with a household thermometer. As we use an open test bench setup consideration should be given to the fact that the temperatures would likely increase further in a closed case environment.
The GTX275 produced a decidedly average 50c when idling in windows but when put under a serious amount of load using the Furmark stability and benchmarking tool, the graphics card rocketted to almost 90c! Consider that this is in an 'open air' environment, temperatures could feasibly hit 100c in a closed case. Not good at all if you are in the middle of a firefight on your favourite FPS shooter. Intrigued by this blistering temp, I decided to strip the card of it's stock gunk and replace with my last smidge of Arctic Silver 5 and then tested again.
Temps came down to 82c under load conditions with the new paste which is still not great but a little better. Increasing the fan speed manually did drop temps further but I would have preffered the card to have controlled it's temperatures better by itself. A good point to come out of this is that the fan is silent when idle and not ear piercingly loud even when set manually to 100%. It's a shame then that the card did not take advantage of this by adjusting the fan speed accordingly when temperatures became uncomfortable. It's fair to suggest that this may just be our sample that runs hot as similar results were found with the GTX280. I know I've personally had 3 and all produced wildly varying temps.
For our overclocking tests I used the RivaTuner utility which worked perfectly with our setup. To test stability I ran 3D Mark 06 and a few runs of Call of Duty 4.
Overclocking the card was terrible until I had figured out that it was the temperatures that were locking the graphics card up and not a voltage limitation. With the stock paste substituted and the fan speed set to a bareable 70% I got much better success, achieving a stable 750MHz on the GPU core, 1664MHz shader speed and a reasonable 1248MHz on the memory. It should be noted that in stock format I was not able to progress past 680MHz on the core, most likely a result to the temperature barrier.
The results of this overclock can be seen below:
Replacing the stock gunk and spending a little time with rivatuner resulted in a worthwhile increase in frames per second at all resolutions. However, you will need to consider if this overclock and increase in FPS is worth voiding the warranty should your sample need the paste replacing.