Just to shut you Intel Lovers up!

AMD Dual wanks over Intel:

Athlon 64 Takes the Top Spot

Whenever we review a new AMD or Intel processor, we run some standard game benchmarks and invariably find that Athlon 64 processors soundly outperform Pentium 4 processors. You can find recent examples in our reviews of the Pentium 4 820 and Athlon 64 FX-57. These gaming benchmarks are of the standard timedemo variety, either playing back static frames as fast as possible, or flying through game levels on scripted paths. This article tried to answer the question: What happens during real gameplay? Does Intel catch up, or is AMD still the superior gaming processor?

Pentium 4 Average FPS Athlon 64 Average FPS Pentium 4 Time Under Threshold Athlon 64 Time Under Threshold

Half-Life 2 59 69 28% 6%

F.E.A.R. Demo 64 63 19% 21%

Battlefield 2 57 64 28% 6%

Dawn of War 42 51 18% 14%

Dungeon Siege 2 41 50 69% 27%

World of Warcraft 61 68 69% 10%

Average 54 61 31% 14%

The results speak for themselves. The average frame rate across all six games for the Athlon 64 system is 61fps, while the Pentium 4 averaged 54fps. That’s a 13% difference—not tiny, but not large enough to bowl us over. What is more important, we feel, is how often a game runs slowly enough that you can feel it. This methodology is consistent with the one used by a new performance analysis tool in the works at Intel. We picked arbitrary performance thresholds, but these are numbers based on years of game playing experience. We picked frame rates at which you actually notice an impact on how the game feels, not the absolute minimum required to play and enjoy a game. This is where the Athlon 64 really kicks the Pentium 4 in the teeth. Our P4 system spent almost a third of the time, across all games, beneath our target minimum FPS. The Athlon 64 system, on the other hand, spent only 14% of its time there. This is a difference of a whopping 121%!

Clearly, the results we get from timedemo-style benchmarks in our processor reviews aren’t far off the mark. If anything, you could say they’re kind to Intel. By focusing on average frame rate in the playback of pre-recorded or scripted demos, we find AMD processors are typically 15-25% faster in gaming scenarios. Focusing on the amount of time spent beneath a minimum FPS threshold makes the situation look far worse for Intel, as they spend more than twice as much time beneath the limit.

Is this a better way to benchmark? Possibly, for some scenarios, but it’s too impractical to use. It’s impossible to play the game the same way twice, and even if you do, there’s no way to make the dynamic systems (AI, physics, etc.) behave the same way each time. Each benchmark run is an inconsistent play experience, and it’s impossible to eliminate enough variables even when running them multiple times. What’s more, taking the average of multiple runs will often hide some of the warts you’re there to unveil. It’s completely impossible to expect other analysts to be able to approach even a modest amount of consistency, and you can imagine that running these benchmarks again a month or two from now, when a new CPU hits the market, would yield somewhat different results.

http://www.extremetech.com/article2/0,1697,1854741,00.asp