Related Articles

RightMark Memory Analyzer

Here the dual memory channels of the nForce 4 shine. The 670 gets the best results, closely followed by the 560. There is a healthy 20% bandwidth advantage for the 670 over the best Athlon 64 result.

Ok, let's face it: there are definite advantages to moving the memory controller onto the processor. This test clearly illustrates the large gains to be had by the processor directly controlling the memory; the read latency figures do not lie. The best Athlon 64 figures show a commanding 64% faster read latency than the best 560 results. The 670 trails very closely behind the 560.

The 670 is the king here. It is very interesting that the maximum figures for the Athlon 64s are relatively close (within 8%) of the Intel figures, while the average figures fall far behind with a good 40%+ loss. Given the low read latency of the Athlon 64s, this is very puzzling.

The Athlon 64s are approximately 50% faster for the maximum Real Write tests. On the average readings, they are running neck and neck with their Intel counterparts. This is also puzzling. I am beginning to wonder if the RightMark average numbers may not be affected by the memory access patterns used by the benchmark.