integrated FSB
 

I'm just curious if integrated FSB is actually running at the speed it is supposed to. Is it sort of like 64-bit where the technology isn't really supported yet? Or is something like FSB a feature that doesn't require something specially formatted or designed to be used?

Another question I have is how do CPUs effectivly operate with such low FSB? My friend has an AMD 2800 and runs HL2 fine. I don't really get how this works since his FSB is 333MHz. Does the FSB not effect the overall speed as I'm making it out to? If the FSB really played a big role in overall output of the cpu, then wouldn't 333Mhz CPU FSBs be very slow computers?

FSB is not all that makes the CPU "fast" as Intel clearly showed us in the past. The FPU's (Floting Point Unit) efficiency is what makes the CPU really work well. AMD Kicked Intels ass there for a while (early P4's) with half the mhz and FSB because it's FPU was more efficent. In fact Intels old P3 chips were beating out the first P4's even though the P4's were 1000mhz faster and twice the FSB speed.

333mhz FSB is really 133FSB duel channel ext... The intel 800mhz FSB is really 200mhz quad channel.... ext....

Well that helps :)

Thanks a bunch!

just a minor correction
333mhz FSB is realy 166FSB duel channel ext :D

Yes, I stand corrected.... Which is bad because that is the FSB my main rig uses. :D

Well, it's really more complicated, but jaydee certainly isn't wrong.

FSB helps feed the CPU. But, if the CPU is working on the same data for a while, the FSB is less significant. If the CPU is doing minimal processing on a lot of data, then the FSB is more significant. Also, different kinds of processing uses different parts of the CPU. So, really, many different parts of the CPU (FSB, multiplier, address decoder, FP unit, etc...) can be the weak link.

It really depends upon the program(s) you are running. ID games run better on Intel. Ray tracing runs better on AMD.

Both AMD and Intel use clock doubling techniques on their FSB to increase bandwidth while not really decreasing latency. So, that works better on data processing but not as good in branch intensive code like A.I. Anyways, if this seems complicated, just think of it as nice magic, because it's the reality is more complicated than either Jaydee or I understand, much less explain. (I hope I am not underestimating Jaydee's EE knowledge. If so, please excuse me!)

Not at all, your far more educated in EE than I am.

Its the other way around. Q3 and D3 both favored the Athlon64.

http://www.anandtech.com/cpuchipsets...spx?i=2149&p=7
http://www.tbreak.com/reviews/articl...6&pagenumber=4
http://www.hothardware.com/printarti...?articleid=701

And one inline since I can't find the artical the link is from (hopefully Anandtech won't mind):

http://images.anandtech.com/graphs/i...01117/5451.png

The P4 really gets destroyed in Doom3. It fastest Prescott is barely beating the slowest A64.

Thank you! I thought it was the other way around. Maybe I just remembered the high-end P4's were doing better than the Socket-A CPUs? Well, in any case, the AMD chips are doing great these days. But, it probably isn't due to one single factor like changing the FSB.

Nor, are great CPU temps (usually!) only gotten by changing one thing like the radiator.

I'm gunna go cry, I really have lost any concept of how CPUs work now...You'd think a 3.8 Ghz CPU would out-preform a 2.6....So confused! thanks for the infor nayway guys