Tighten up my Ballistix

[mysterfix]

If your running stable in Windows and 3D apps then I'd say don't worry about memtest. Prime95 is a lot more stressful than memtest and many people have proven that even if your ram passes memtest that doesn't mean that it will run stable in 3D modes such as games and benches.

Edit: I have the same issue with everest, I'd recommend running it multiple times per setting change.

 

[jason4207]

Gotcha!

One more thing...Everest is giving me the following results:

read: 8686
write 2928
latency: 5.8ns

I saw an eva2000 post (I know I shouldn't compare to him), and his write score was a lot closer to his read score. Should this be cause for alarm ,or are my scores typical?

Edit: P95 says:

small fft: RAM not tested much
large fft: some RAM tested
Blend: lots of RAM tested (makes me think this is the best RAM test)

But Orthos says:
small fft: stress CPU
large fft: stress some RAM (makes me think this is the best RAM test)
Blend: stress CPU and RAM

It's a little funny how they word it, but I'll take your word that large fft is the way to go for testing RAM

 

[jason4207]

I'm running 5:6 1080MHz 2.2v 5-5-4-4-10 4-25-7-10-11-3-8-3-5-2N now. It seems to be stable so far. Only minimal everest, 3DMark06, and SuperPi increases, though.

I'll probably stay 1:1.

Edit: 4:5 1125MHz was giving me P95 errors, but I didn't try to loosen the timings or increase the voltage yet. I think 1:1 is the way to go anyway.