Testing on that low resolution without filter higher fps values gain when there wasn't any action in the game is just plain stupid. If you use a test on that low resolution to evaluate which processor that is best than you don't know how these processors behave.
Core 2 has a big L2 cache (6MB) with 15 cycles delay.
Phenom II has a smaller L2 cache (512 KB) with 15 cycles delay, L3 cache is 6 MB with close to 40 cycles delay and the L3 cache runs at lower frequency compared to Core 2 so maybe the L2 cache on Core 2 is about 4 or 5 times faster accessing the L2 cache compared to when Phenom access the L3 cache. L3 cache has a lot of advantages running multiple processes etc but just feeding one core with data and the data isn't shared or fragmented then the Core 2 L2 cache is much faster compared to the L3 cache on Phenom.
Games love cache, if there isn't any action in the game the cache will very much decide the speed of the processor. Areas with little activity is much more common in games compared to areas with high activity. If there isn't anything else that slows the processor like a graphic card that needs to redraw a complex picture at high resolutions the Core 2 will gain a LOT of FPS values when there ISN'T any action.
Is this a good measurement for how good the processor is in games? No it isn't.
What you should try to do to really test the processor for how good it is in games is to remove all those areas that isn't important (when there isn't any action), it isn't there the game will lag. And focus on areas where the processor needs to work hard (when there is action in the game).
What Kyle did in his test was to focus on areas that isn't important and used that to conclude how good the processor was. Either he don't understand hardware or he wanted to make AMD look bad.
Originally Posted by panfist
From the HardOCP article, which you apparently didn't read:
Bookmarks