Benchmarks I
It's feature-laden, now let's see how it performs. First up is SiSoft SANDRA's buffered benchmark. Buffering shows us the potential bandwidth for any given memory speed.
Quite expectedly, running memory synchronous to the CPU's FSB gives us the lowest scores. DDR266 is the memory speed that the i845E chipset officially supports, so in that context, it is the slowest here. Running memory outside official specification with our 3:4 FSB:DRAM trick gives us the highest scores. We're closing in on 3GB/s with the processor still at its native 133FSB.
We gain an impressive, theoretical 36% of bandwidth by changing memory speeds independent of the FSB. How does this translate into performance ?.
Pifast, our first practical test, simply calculates the constant Pi to the desired number of decimal places. I've chosen 10 million using the fastest method possible. Memory bandwidth has historically played a major part in this benchmark.
The 6-second difference between the MAX3's time, when using either DDR266 or DDR356, is quite staggering considering that the CPU has remained constant at ~ 2800MHz. The MAX2 manages to eek ahead of its stablemate by the smallest of margins. Going strictly by specifications, the i845E chipset is hampered by the DDR266 memory speed. Artificially boosting that speed pays great dividends. Having said that, timings are absolutely vital in this benchmark. Relaxing the MAX2's timings from 2-5-2-2 to 2-6-3-3 increases the time from 76.05 seconds to 78.55 seconds. You need superb RAM to bring out the best in this motherboard.
Next we'll turn our attention to MP3 encoding. We're benchmarking by encoding a 638MB custom WAV file (Moby's Play album, incidentally) into 192kb/s MP3 using the LAME 3.92 encoder and Razor-Lame 1.15 front-end.
The inbuilt timer calculates to the nearest second. With that in mind, we don't see a difference between the results even though the system memory has been altered. This benchmark is all about the CPU.
