If you run your FSB and Memory bus at the same speed, there are no latency penalties for the chipset having to syncronize the two busses during data transfers... it's one to one and that's good. The thing to remember about DDR is that the data throughput is doubled so while the bus runs at only 133 MHz. it transfers data at the equivalent of 266 MHz.
If you have your FSB at 133 MHz. like many Athlon's, but your memory at DDR333 which means the memory bus is running at 166 MHz. every time there is a data transfer, the chipset has to manage the two different frequencies. Not only that, but it says that your memory data throughput is greater than the data throughput to the CPU, so you get into diminishing returns when you have more memory bandwidth than FSB bandwidth.
Lately, Intel has held a commanding lead in their data throughput to the CPU with their Quad pumped FSB, which is four times the actual frequency. So my Northwood P4 runs it's FSB at 133 MHz, but it transfers data at an equivalent of 533 MHz. which is largely why their recent P4's have done so much better than they did originally. (Of course, now the new P4's use a 200 MHz. FSB quad pumped to 800 MHz!)
As for those 2-3-3-5 numbers, those are the various memory latency settings you can control in your BIOS and they really have nothing to do with the sync of your busses other than the faster you put your memory bus, the greater you need to make these memory latencies, depending of course on the rated speed of your memory.
Scout
700 Mflops in SETI!
Facebook
Twitter
Instagram