im using socket 754 sempron 2500 @ 2.1ghz, the supported multiplier for the hypertransport is 1-5X. therefore if i overclock to 2.1ghz, i have to raise the hypertransport to 300MHz, the multiplier which i can savely boot into windows is 2.5X. which is a total of 750MHz. i benchmark it using super PI, pcmark 04, 3d mark 05. then i lower the multiplier to 1X (total 300MHz hypertransport speed) i benchmark again using the same utitlities, the marks are the same. i try many times and still the same result between 750MHz HT and 300MHz HT.

so the question is, the HT multiplier is nothing? HT is just a marketing strategy from AMD? maybe the upcoming AM3 with HT 3.0 will perform the same as AM2 HT 2.0? i guess so.

It seems that you do not have any idea what Hypertransport really is. HT is not a marketing strategy, and higher HT speed was never meant to show performance increase. It's the bus that communicates cpu with the chipset. The chipset is the pci-e controller, as well as other peripherals such as SATA, IDE, etc.

The reason why higher HT speed does not increase performance is because it does not have anything to do with the communication between cpu and memory simply because of the integrated memory controller inside the cpu.

The reason why AM3 will incorporate faster HT bus (3.0) is to allow faster connection between the peripheral to multicore platform.

soulfly, i still dont understand how the HT could help in anyway the communication between peripherals of a PC. how does that compare to non HT systems like those of intels?

Hypertransport is just a substitute to the conventional front side bus.

From what I understand, with hypertransport, cpu can communicate directly with PCI-e/PCI interface... something like that lar.

With FSB meanwhile the communication between cpu and PCI-e/PCI must get through the northbrige.