im confused...isnt a higher fsb better?

[cntw82oc]

last night i used sandra to test my system....i tested it at 1362 (12.5*109) & 1359 (9.5 * 143)

sandra gave me considerably higher scores @ 12.5*109 than @ 9.5*143......whats up with that? can someone explain?

[seabhac]

Higher FSB=higher stress on memory and cards.
Higher CPU multiplier=higher stress on cpu.

So it's really relative. If you can get higher speeds on your FSB thats the way to go. Just bear in mind it stresses out the cards and ram. You probably have more invested in sound, video, mem, etc. than you do in your chip.

Higher FSB improves memory performance though.

[cntw82oc]

thats what i dont understand...i can get to the same speed both ways (higher mult. and lower fsb...and vice versa) so why arent i getting better benchmarks at the higher fsb?

[William]

is your ram optimized for the higher fsb, if not ,that could easily account for the difficulties.

[Thelemac]

cntw82oc (Apr 17, 2001 10:54 a.m.):
last night i used sandra to test my system....i tested it at 1362 (12.5*109) & 1359 (9.5 * 143)

sandra gave me considerably higher scores @ 12.5*109 than @ 9.5*143......whats up with that? can someone explain?

What test were you running? And which results are you reporting? There are so many different tests in Sandra that you really can't just tell us one number...

Also, three points really isn't all that different. If you're doing the cpu test, then I would say that's probably about right...if you really want a good idea of how your system runs, you should run the tests five or six times and average the scores, cause the tests will run differently every time.

[Fiz]

It may be because at 109 the PCI bus is higher than it is at 143, at 109 i think its at 36 or 37 MHZ, at 143 I think its at like 34 or 35. That could cause it.

Fiz

[diehrd]

The systems overall performance will be greater with a higher FSB,and you must average the Sandra test scores,KEEP in mind a lower clock and higher FSB will result in a lower cpu score and a higher memory score.

[kat]

when you run a high fsb these are the benchmark test`s that will show a great improvement ;D

[Sirfin]

Took me awhile to figuire out which was better for me too.

1st clocked at 114x12.5 (110x36) =1430 mhz
2nd clocked at 130x11 (130x32) =1432 mhz

Sandra Results:
CPU bench clock 1. 3922 & 1913
CPU bench clock 2. 3931 & 1932
MM bench clock 1. 7921 & 9653
MM bench clock 2. 7918 & 9633
Mem bench clock 1. 399 & 432
Mem bench clock 2490 & 531

3dmark 2001 clock 1: 2316 clock 2: 2355

For me was a dramatic improvement to run fsb at 130x 32 -vs- 110x36 and drop my multiplier to 11 -vs- 12.5 Was pretty pumped to get my pc-100 ram up to 130 mhz. :-)

Hope that helps,

Sirfin~

[asmodean]

I always stick to the higher FSB.
My memory works at 142MHz/CAS2... Shees. Specs for Kingmax is 133/CAS3. Wouldn't even mention that Eudar, it's pigcheap. But hey, it WORKS.

Results for Sandra Mem: (about)
Int 595 & FPU 675

I keep Kingmax at Turbo, Eudar at Fast.

[Hoot]

It's a balancing act...
What I have found is if the multipliers are close, the one with the higher FSB is the better choice. If you have to lower your multiplier considerably, IE 12.5->9.5, that's too much of a trade-off. Also, you have to look at how close each one is to the unstable threshold. For those of us who don't have a 1.33 cpu, there is usually a multiplier setting that regardless of FSB, we can't get over. Likewise, for our particular memory and add-on cards, there is usually an FSB we simply can not get over. The trick lies in easing up to that threshold and varying the other parameter with stability and temp in mind. That juggling and balancing act is what makes OC'ing fun.

Hoot