What would provide better results?

[supastar1568]

hey folks,

finally got my system stable at a nice 3.8Ghz. I have tested stability for two different conifgurations. I can run my system at 3.8Ghz, 423 FSB x 9, with my ram running a divider of 6:5 at 507 MHz (stock speed is 533). Or I can run my system at 475 FSB x 8 with my ram running at 1:1, 475 of course.

In theory, which should provide better results or is the better configuration?

Both ways are 100% stable (24hr small FFT, 24 hour blend, 8 hr memtest)

 

[cubeless]

download prime95 and do the benchmark...

in real life i don't think u will see a big difference between the two... 1:1 @ 475 should be a little easier on your hardware... from what i've seen you don't lose anything by using a divider, tho... so technically 507 is better (at the same timings)...

or u can push up the volts on the mem and get to 533... ;p...

them intel chips like more bandwidth, the timings don't seem to be as important as with the amd chips... loosen up the timings and make a run for 1066!!!

 

[bunnyfubbles]

I'd stick to 1:1 and tighten the timings as much as I could, but that's just my personal preference, I don't think you're going to get a real difference either way. Each might have an advantage over the other if you're looking to optimize your benchmark scores for something particular, but in the end it won't really matter when it comes to the meat and potatoes performance that you can actually eat and fill up on...

 

[supastar1568]

Originally posted by: bunnyfubbles
I'd stick to 1:1 and tighten the timings as much as I could, but that's just my personal preference, I don't think you're going to get a real difference either way. Each might have an advantage over the other if you're looking to optimize your benchmark scores for something particular, but in the end it won't really matter when it comes to the meat and potatoes performance that you can actually eat and fill up on...

Yeah, I agree, think I'm staying with the 1:1 ratio. I ran the superPI and Prime95 benchmark tests. The results were very close, with the slight edge going to the 1:1 ratio at 475.