Originally Posted by pshootr
Very interesting point. As a new IBT user, I have not been impressed with it even from the get go. It just seemed so hit or miss to me. I could run 12 hours of Small FFT with stock ram speeds, then fail IBT. Personally I can not recommend it for estimating stability.
Well, are we on the Stable Question again?
Note: Small FFT uses lesser RAM than IBT AVX Very High. A fair comparison should be Prime Blend.
Originally Posted by Chris635
Are you guys noticing performance increase with a 2600 mhz CPU/NB. I have run 2600 and I haven't seen any difference than my current 2432 (my ram is clocked to 2432 mhz). I am curious.
Stable CPU-NB will prove a little more MHz and cut off a little bit of latency for the RAM and L3 cache. It's a bit trick if it's unstable. Seeing you have your RAM at 2400+ MHz, it should post better scores on AIDA 64 at 2600+ on same timings.
Lower RAM speeds than 2133 wont benefit much.
Originally Posted by xKrNMBoYx
No clue, my RAM is only at 2133MHz. I heard the sweet spot is to have both HT and NB the same for FX chips. I may downclock the NB/HT to 2400MHz when I RMA my RAM that actually work at 2400MHz but until then I'll keep it at 2600MHz. I just don't like the idea of running HT lower than 2600MHz when that is the stock speeds. I guess benchmarks could be one way to see the difference but probably little difference in real world performance/situations
So far I've gotten 2600MHz stable with CPU/NB Voltage set to 1.275V in the BIOS while testing IBT VH 20 Runs. Is there any other program that will test for NB stability better than IBT/P95? I'm about to test 2600MHz on 1.25V CPU/NB Voltage.
OCCT Custom with defined amount of RAM tested. With AVX instruction set.