someone teach me something

I am finding that using multipliers allows higher speeds with less vcore than a comparable MHz arrived at using higher fsb. Can someone tell me in plain English why that is? Thanks. My experience so far: 192 x 11 takes 1.725v to run stable and it idles at about 41C; 169 x 12.5 only uses 1.675 and idles at 38C.

good question...I would also like to be enlightened..

SP

hmn let me take a guess, because when you pump up FSB everythings working faster, (the pipelines to/from chip)
whereas the chip itself endures the brunt of the O/C when you get it higher with multi's?

my two ¢

That makes sense. So I am actually running faster with the higher fsb even though the MHz are the same?

I found that this same thing was true with my 1800+ Tbred A and Abit KX7 mobo. I could run 12*150=1800 with 1.60 Vc, but 10*180=1800 required 1.65 Vc. I think it must have something to do with running higher FSB speeds, even though the CPU is running the same or less MHz.

And YES, running 192*11 will be faster than running 169*12.5 due to the higher FSB speed. The higher FSB will greatly improve your memory bandwidth and make your computer feel a little faster. Try running the Sandra Memory Benchmark and you will see the difference.

bk94si said:

I am finding that using multipliers allows higher speeds with less vcore than a comparable MHz arrived at using higher fsb. Can someone tell me in plain English why that is? Thanks. My experience so far: 192 x 11 takes 1.725v to run stable and it idles at about 41C; 169 x 12.5 only uses 1.675 and idles at 38C.

Click to expand...

In general, when any part (such as CPU clock, or FSB clock) of a chip runs faster beyond certian limit, it would require higher voltage to go faster. This is because higher voltage can supply more current to the transistors involved. And higher current means the few10 millions capacitors inside the chip can be charged and discharged faster to overcome any slowness in some logic paths inside the chip that might lead to system instability. Of course, higher voltage (V) means higher power dissipation (proportional to V^2 frequency) and higher chip temperature that we all know about.

Having said that, the higher voltage to run higher FSB (at the samel CPU speed) is not waated. Since the FSB is increased from 169 MHz to 192 MHz, which is 23 MHz faster, with a higher Vcore of 50 mV (0.05V). The memory bandwidth is increased by about 8 x 2 x 23 = 368 MB/s. That is more data per sec can be transferred to and from the system memory and the CPU FSB interface and then CPU internal. One would notice a considerable gain in system performance, especially applications such as 3D graphics, games, video streaming, programs with huge data (such as large scientific matrices), .... One would see a big difference in 3DMark 01 benchmark score with the 23 MHz faster FSB.

Thanks for all the input. I guess what it really says is that everyone needs to stop fixating on how many MHz their computer is running at.

bk94si said:

I am finding that using multipliers allows higher speeds with less vcore than a comparable MHz arrived at using higher fsb. Can someone tell me in plain English why that is? Thanks. My experience so far: 192 x 11 takes 1.725v to run stable and it idles at about 41C; 169 x 12.5 only uses 1.675 and idles at 38C.

Click to expand...

I would answer: more occilations per time unit (up and downs)-> more voltage to get clear (readable) occilations.

For anyone that doesnt know: You system clock speed is made multipling the front side bus by the clock multiplier. For a very very lose comparison lets compare it to a car. You add nitrious oxide to a 6 cylinder car it has a higher combustion rate and moves faster. That would be the same as upping the front side bus. You are taking the existing settings and stressing them past what they were supposed to be stressed at. By upping the front side bus you pumping NOS into your system. On the other hand you take the same car and take out the v6 engine and put in something like a v10 engine and get the higher power as you would by pumping the nos into the V6 engine but the V10 engine was meant to be able to handle the stress. Its essentially the same as changing the clock multiplier. Your keeping everything the same but adding in a new part, well sort of. When you change the clock multipliar in the system its like putting in a new factory CPU thats rated higher. IT like putting a bigger engine in your computer rather then putting further stress in the original engine. The new engine is meant to handle that extra stress,. hence the lower voltage and heat. When you chagne the front side bus on your computer its like adding NOS to it and stressing it further. But when you change the clock multipliar its like putting a new engine in the computer. Your not makeing the processor faster, your essentialy changing it all together. I know this is confusing, and there are tons of spelling mistakes, I wrote this up quickly. You got further questions or anything email me and I can take the time to write it up and explain it better. [email protected]

all that talk about capacitators and about a million acronyms has gotten me confused
________
MERCEDES-BENZ S-CLASS

dalilman52 said:

all that talk about capacitators and about a million acronyms has gotten me confused

Click to expand...

Probably you refer to my explanation, let me try again. Sorry for too much technical details.

CPU has two mainly parts:
- the computation part whose transistors are driven by a CPU clock (around 2000-2500 MHz for AMD CPU). It is mainly for doing calculations.
- the FSB part whose transistors are driven by a FSB clock (around 133 - 166 - 200+ MHz for AMD CPU). It is mainly for transferring data in and out of the CPU to the rest of the system, namely the northbridge and then system memory and video card, ...

When the FSB is running faster (beyond certain limit), it requires more voltage, and so more electrical power and hence dissiplating more heat. This is generally true for any chips.
The FSB running faster from 169 MHz to 192 MHz as in the previous example, means the CPU can transfer and receive data from the external system memory by much more bytes per second, even it is doing the computations at the same speed, in both cases. 1 byte = 8 bits of data. To be exact, it was about 368 MB/s more for FSB data transfer.

Hope this helps better. Further questions, pls post.


Next two links explain some basic terms, one may skip them.

What is cycle time and frequency

Frequency, clock, period of synchronous operations, latency