World Exclusive Mushkin UltimateFX GeForce GTX 295 Review

Posted by: James Clewer (gt_junkie) on: 09/08/2009 05:39 PM [ Print | 0 comment(s) ]

Testing

The Mushkin GTX 295 will be compared alongside a couple of standard GTX 260's (65nm - 192 stream processor model) run both individually and in dual card SLi (both cards are identical to the reference NVidia design running at standard clocks) as well as two Zotac cards comprising of the highly overclocked ‘GTX 285 AMP! Edition’ and a standard clocked dual PCB GTX 295.

Full spec of the test PC is:

CPU: AMD Phenom II 720 Tri-core @ 3.4ghz
Motherboard: ASUS M3N-HT Deluxe Heatpipe
Memory: 8gb (4x 2gb) OCZ Platinum PC8000 DDR2
HDD: 320gb Samsung Spinpoint F1
Operating system: Vista 64 Business
GPU: 2x GTX 260's, Zotac GTX 285 AMP! Edition, Zotac GTX 295 and the Mushkin GTX 295
PSU: Tagan Piperock 1100w (modular)
Cooling: Custom watercooling for CPU and motherboard chipset - GPU's all use their original stock cooling
Case: Mountain Mods - H2g0 enclosure

The UltimateFX GeForce GTX 295 in place.

Initial testing was compiled using Futuremark's two most recent GPU benchmarking programs - the older 3DMark06 and their newest 3DMark Vantage. These were the newest patched versions at the time of the review.
(Note: In the case of 3DMark Vantage PhysX was enabled in the 'Nvidia Control Panel' since it was utilised in the test.)

The GTX 260 and GTX 285 GPU's were tested using NVidia's GeForce driver set - 182.50. The Zotac GTX 295 was tested using NVidea's GeForce 186.16 driver set.
The Mushkin GTX 295 was tested using NVidia's newest 190.38 driver set.

Note: As in my previous GTX 295 review i did try both the 186.18 and 190.38 driver sets with the Mushkin GTX 295 to test for any performance changes. On this occasion i noticed a slight increase in performance virtually across the board. There were a couple of blips but this could be down to natural variances in the testing.

The operating temperature of this newer single PCB'd GTX 295 was cooler than that of the dual PCB version of the card. When idling each core was 5c cooler but more astonishingly the full load temperatures were 17-18c cooler on each core. A massive difference i'm sure you'll agree - it's all the more impressive when you consider that the production process (and consequently power usage) is identical to the 'hot running' dual PCB version. It seems that the split heatsinks may aid the cooling of the card by some considerable amount. Full load temps were achieved by looping the Far Cry 2 long Ranch demo and ambient temperatures only fluctuated by 3 degrees at most.