High FSB vs High Mult

  • Thread starter Deleted member 143938
  • Start date
D

Deleted member 143938

Guest
Why do people keep setting a very low FSB and a high multi? I usually see people do something like 220x12 instead of something like 300+x9 let's say... why? I'm sure their RAM can handle >220
 
In my opinion the more the better, pushi it as fast as it can go higher fsb translates into higher mem frequency which in turn translates into an over all faster system. it all just depends on your components. check sig.
 
My CPU won't go above 225 fsb, no matter what I try. thus, I am stuck at 14x225 (fx-55 San Diego). That's still 550Mhz over stock, not bad for $139
 
b/c the processors don't need the extra memory bandwidth from a high FSB. It translates to practically zero improvement for real world applications, the only difference you'll really see if a memory benchamrk!

Look at the difference between a 939 & AM2 system. At the same FSB, the AM2 has a HUGE memory bandwidth improvement, but actual real world performance is nearly identical.

I experimented around with a bunch of different FSB x multi combinations on my Core 2 system and the difference between best & worse was less than 1% and some things actually performed better at the higher multiplier setting for me. I'm sure it had to do with having to change memory ratio & timings at different FSB's.

9x334 vs 8x375 vs 7x430

7x was fastest, just slightly
9x was good
8x was slightly slower than 9x

Anyway, I assume you're talking about an AMD system and with its IMC it really has all the low latency bandwidth it needs. Granted more might help a little, but it could also decrease stability, since the motherboard & HT might get a little strained.
 
It depends on the motherboard as well. I would love to have a FSB over 300, but motherboard limits me to 260-sih... therefore I have to run a high multi to get the overclock I want.
 
MrFace said:
Athlon 64 is about mhz, not FSB
Click to expand...

So is Core 2.

Cranking up the FSB only helps a lot if the processor is bandwidth starved. Neither (core 2 or A64) is bandwith starved like some earlier processors!

Athlon 64 has an integrated memory controller so it gets plenty bandwidth.

& Core 2 has a great prefetcher & large L2 cache which gives it enough bandwidth.

Both have dual channel memory controllers too.
 
It also puts less strain on your other components that are associated with FSB frequencies(ie. your mobo and your ram) and focuses the ability to overclock on your CPU directly. It also lets you get away with relatively "cheaper" components. For example, you might only need to get DDR2 800 ram instead of DDR2 1000 to reach the same processor speed.
 
A higher FSB puts extra stress on both the chipset and the RAM, which means that other components will have a higher chance of limiting the CPU. Using a higher multiplier allows you to keep the FSB at a more reasonable level, which allows the RAM and chipset to be under less stress while the CPU reaches the same or even a higher speed.
 
400x8 is slightly faster in than 356x9 for my E6600 (atleast in 3Dmark06 the CPU score was 40-50 higher), which I found odd because a 400 FSB, my ram runs at 800Mhz as opposed to 356FSb where it ran at 890Mhz.
 
40-50 higher?

Did you re-run that a few times? 3Dmark varies from run to run by a few points.
 
Arcygenical said:
40-50 higher?

Did you re-run that a few times? 3Dmark varies from run to run by a few points.
Click to expand...

Yup... At 400x8 I score around 2800-2810, and a 356x9 I score around 2760-2770

I might do some re-runs to make sure I wasn't running anything else that may having been using up some resources.
 
400 FSB & 333 FSB are optimal settings b/c of the FSB strap.

Thats probably why 333 (with 2.5 memory ratio) worked best for me.
 
You must log in or register to reply here.
Back
Top