While we feel the truth might lie somewhere in the middle, it is time to take another look at the Nvidia GeForce GTX 480/470 vs. ATI Radeon HD 5870/5850 comparison.
The GeForce 400 series, particularly the GeForce GTX 480, is a serious piece of hardware that at least on paper should leave the Radeon HD 5870 dead in the water. Our first go at this board revealed that while it was priced ~20% higher than the Radeon HD 5870, on average we only saw a 16% performance increase based on the dozen games included in our benchmarking phase. Perhaps more relevant than that, we found that when placed under full load the GeForce GTX 480 consumed ~30% more power, while at the same time it attempted to send us deaf.
When wrapping up our original GeForce GTX 480 review two months ago we had this to say...
The GeForce GTX 480 is fast, but given the extra time Nvidia had to work on the card and tweak it to perfection, we would have at least expected to do without the heat/power compromises. As drivers mature these Fermi-based graphics cards will likely become even faster, but we may also suggest Nvidia revises its pricing strategy, which could make or break sales once the cards become available next month.
It would appear that time is now as Nvidia released the 257 beta driver revision a couple of weeks ago. Nvidia claims these drivers provide significant performance improvements across the board for the GeForce GTX 400 series. In the 8 weeks since releasing the GeForce GTX 400 series AMD has also polished their Catalyst drivers and the current 10.5 version seems very solid.
Another reason why we wish to revisit this high-end graphics card battle is to settle a belief of green team fans that the GeForce GTX 480 is far superior to the Radeon HD 5870 when measuring minimum frames per second. This is an interesting argument as the average fps results did not suggest a huge difference in minimum frame rate performance when we last tested, though it could be possible, and therefore we wanted to find out if this was true or not.
However, simply showing the flat minimum frame rates is not enough considering that this could be the result of something else going on within the system that causes a split second drop in performance. Therefore we have recorded a frames per second timeline to reveal just how often the performance drops and for how long. Shiny new graphs coming your way, read on...