Benchmarks: ScienceMark, HEXUS.pifast and Cryptography
Running slightly castrated, our DDR2 RAM more or less equalled the bandwidth of the Socket 939, DDR1 RS482-M. Had it been running at full whack, it would no doubt have had an advantage in this particular test.
The DDR2's increased data rate isn't enough to overcome its latency. The gap might have been smaller at DDR2-533 speeds. Of course, these memory issues aren't the fault of the RS485M-M's as such, rather the CPU in use, but we chose a CPU that somewhere in the middle as far as AM2 performance options go, to try to deliver more accessible benchmark results.
Crunching out 10M places of pi takes just over a minute, three seconds slower than the same rating of CPU on Socket 939.
In the HEXUS Cryptography benchmark, the difference in result between the two boards isn't significant enough to mention.
Impressions thus far then, show nothing to suggest the board's holding anything up, rather the DDR2 RAM.
