Benchmarks I
We're changing system memory speeds when raising the XP2500's speed, and we also have a number of processors that use differing FSBs and two platforms. A look at SiSoft SANDRA's unbuffered memory benchmark should yield some interesting results.
Running a double-pumped FSB CPU on a 166MHz FSB (333MHz, effectively) is never going to produce high SANDRA benchmarks. That's why the XP2500+ and XP2800+ produce very similar results. The XP3200+, though, stretches its 200MHz FSB legs a little, and the XP2500+ Barton, overclocked to 2332MHz / 212FSB, adds that little extra performance. There was only ever going to be one winner in this test, it has to be said.
Pifast v4.1, on the other hand, our resident Pi calculator to 10m places, has always been a fan of the Bartons' ability to crunch mathematical data. A high clock and memory speed allows the 2332MHz AQXEA Barton to break the 60-second barrier. It takes around 16 seconds out of the standard XP2500's performance. Overclocking does indeed pay.
WAV crunchin' is one of the tests that is used in motherboard reviews. However, the benchmark truly excels when one adds in clock speed. It's plain to see that the overclocked Barton still fails to topple the 3.2GHz P4's time of 2m 45s. Let's not forget we're comparing a CPU that's 8x more expensive than our Barton stalwart.
Pretty good so far.
