Towards the later part of the benchmarks we've dispensed with our older methods and decided to go a little bit more in-depth for you gamers out there. We already know that the GeForce 2 GTS is the performance champion of 3D cards and thus we charged up three fully loaded 1GHz systems to see which was the "Ultimate Quake-platform".

With WinBench 2000 having been released, we got straight into the "Processor Test". It uses the "NULL" device when testing/simulating T&L routines of a 3D application. This puts a CPU's floating-point unit to work determining CPU speed and eliminating the video card as a bottleneck.

WinBench 2000 runs an "infinite renderer" measuring the performance of the CPU functioning alone to work through the various WinBench scenes. Where all of the other synthetic applications used later on in the review measure the FPU and integer units directly, the WinBench 2000 processor test puts stress on the cache. This has been where the Athlon "Classic" has suffered the most in the past when pitted against a similar frequency Pentium III "Coppermine" and its full-speed L2 cache. Now that the specs are somewhat even, check out the results in our graph below.

From this we can conclude that a full speed (smaller) L2 cache built onto the Athlon "Thunderbird's" die has allowed it to bridge that old gap somewhat and regain some ground lost to the Coppermine. Need evidence? Just compare the score of 1.67 for the Athlon "Classic" 1GHz to that of 1.8 for the T-bird. No surprises there, though the Pentium III 1GHz "Coppermine" still edges ahead of the T-bird 1Ghz though at 1.91.

Having spent some time this past week in AMD's Santa Clara facility, we were made aware that AMD's internal benchmarks came out closer to 2.1 (these are the results that will be published in marketing materials). The KT133 and Athlon "Thunderbird" platform provided to us by AMD could not replicate the figure of 2.1 but we will continue to try to do so. Regardless, the gap has closed considerably.