Test SetupOur test system is based on an Intel Core 2 Duo with Intel 975x “bad axe” motherboard, 2Gb of system memory, a fast Western Digital Raptor hard drive and fresh install of Windows XP SP2. While Windows Vista has been released earlier this year, and it has DX10, we’re waiting for games to actually make good use of the new DX10 features to make it worth switching operation system, as well as SP1 to work out the kinks in the system. With OEM’s asking Microsoft to keep XP around a bit longer it seems we’re not the only ones finding a lack of reasons to “upgrade” to a new OS. Our tests were done with the latest NVIDIA Forcedrivers available at the time of writing. For the 7600 GT card it was 94.24, for the 8600 GT’s it was 158.22.
The G84 GPU in all its gloryThe enclosure used for our test is a
Sunbeamtech 3D Storm.
Intel Test Setup |
CPU | Intel Core 2 E6400 @ 2.8Ghz (from CSMSA) |
Cooling | Coolermaster Hyper TX |
Mainboard | Intel 975X Bad Axe (Modded by Piotke) |
Memory | 2 * 1Gb PC6400 OCZ |
Other | Sunbeamtech 3D Storm Antec TruePower Trio! 650W Western Digital 74Gb Raptor SATA HDD
|
Test MethodologyThe low end cards we previously tested did not so good in today’s games, we had higher hopes for the mid range series. Resolutions of 1024x768 proved quite do-able with the new 8600 GT, and in some games we even got to enable Anti Aliasing and Anisotropic Filtering. When comparing performance to the 7600 GT we found that the AF setting of 4x did make a noticeable impact in average FPS, the same setting enabled on the 8600 GT cards did not cause a performance drop, in fact you can safely enable 4xAF on the Geforce 8600 GT without a performance hit, a nice bonus don’t you think?
We threw together a mix/match of game benchmarks to stress the video cards, our first round of testing was concluded when we saw benchmark on the web appear which had the 7600 GT in the lead, over the 8600 GT, without wanting to spoil the ending, the 8600 GT overall is faster, so why did this website report different numbers? We compared the benchmarks used and only had one in common. So we expanded our initial game benchmark selection to include those where the 7600 GT triumphed over the 8600 GT in to the other test.
RPG/Hack&Slash: TES: Oblivion
Racing: Colin McRae DIRT
FPS: Rainbow Six Las Vegas, Prey, FEAR, Ghost Recon 2
RTS: Supreme Commander
Overclocking the 8600 GT cards and 7600 GT
As we’ve seen with the Geforce 8500 GT and the high end 8800GTS/GTX the GPU clocks and Shader clocks are linked, and they increase in steps. The memory can be overclocked 1Mhz at the time but the GPU made jumps of ~10Mhz, while the Shader clocks increased in 54Mhz bumps.
We started with the Sparkle 8600 GT to test the different speed bumps as we increase the GPU speed to the maximum, when we reached a stable overclocked at 666/1458 (evil clock speeds for sure) we continued with the Calibre 8600 GT and got it up to 688/1512. The XFX 7600 GT reached a maximum stable OC on the core of 729Mhz.
Sparkle 8600 GT: GPU 666Mhz (+23%) MEM: 725Mhz (+4%)
Calibre 8600 GT: GPU 688Mhz (+9%) MEM: 910Mhz (+12%)
XFX 7600 GT: GPU 729Mhz (+12%) MEM: 830Mhz (+4%)
The Sparkle impresses most with its 23% overclock on the core; we’ll find out if the increased GPU clock makes a difference in-game. We ran all benchmarks with the video cards at standard as well as overclocked speeds.
The synthetic benchmarks from Futuremark next ->