TheOverclocker Issue 22 | Page 32

Hardware Award GIGABYTE R797TO-3GD RRP: $449.99 | Website: www.gigabyte.com Test Machine Intel Core i7 3770K Corsair Dominator Platinum 2666C10 Gigabyte Z77X-UD4H (F1) CoolerMaster 1.5KW PSU Windows7 64-Bit I f you gazed at the performance results, you’ll have noticed that we have no overclocked results. The reason for that is twofold. First, we had an issue with the particular sample of the graphics card, where it would not overclock much past the reference 1.1GHz clock (isolated to this particular sample only!). Secondly, there would be no point at this juncture. What we’ve noticed is that a lot of the gains to be had with the HD7970 actually stem from the driver releases and 32 The OverClocker Issue 22 | 2012 not overclocking. With the GHz edition cards, what AMD was announcing was essentially a driver update and a BIOS. Well aware of this we took it upon ourselves to see just how much has changed with the latest (at the time) Catalyst 12.11 beta 11 drivers in comparison to the release driver. Before we delve into the performance differences, it’s worth noting that the release driver for the HD7970 was very unstable. We had constant crashes at random intervals and the situation was so dire at some point we considered later drivers. However we managed to get through all the benchmarks in the end. So not only did AMD improve on performance, but we’d wager that most users will appreciate the stability in the long run rather than the higher fps. We would also like to highlight to you that with the older Catalyst 11.12 driver our maximum CPU overclock was a whole 200MHz lower than it was with the Catalyst 12.11 driver. The benchmarks would crash at random times, yet finish at other times (hence the results you see) and we only had total stability at 4600MHz as opposed to 4800MHz. Something to ponder about if you have had an overclocking session with the early drivers that proved to be very frustrating on a CPU you were confident about. On to the performance then as the HD7970 is presented today. This is essentially what the entire point of this review and editorial piece is about. AMD has been optimizing in general but more importantly specific titles and benchmarks. You should note that we did not use any of the CAP updates but simply used the driver package as is. 3DMark03 sees no changes at all, which is expected as neither NVIDIA nor AMD are concerned with a legacy benchmark. Where we do see sizeable changes are in the benchmarks with tessellation more specifically