IT's like I was telling you before, memtest tests JUST them RAM. It doesn't incorporate the CPU 1 bit.
When running prime/S&M the integrated memory controller on the CPU gets used, to a lesser or greater degree.
I run memtest just to see if the RAM can handle the speed, then I integrate in to my CPU overclock and test from there. But then again I find the exact limit of the CPU before I even think about running OC'd RAM.
Yesterday when prime failed after 2.5 hours, I then back tracked and ran s&m memory test and discovered that the memory speed was too high@ 242MHz even though it had passed memtest. I dropped back to the 166 divider which put me at 207.5MHz which passed s&m memory test. I figured that prime would do better after this adjustment, so I turned prime on and went to bed. Well when I awoke 6 hours later, to the sound of that Tornado fan....., I found that core#1 had failed after 1.5 hours, and core#0 was still going. Max temps of 50c(core#0), and 58c on the one that failed.
The settings were 1.5v-253x11.5-RAM@207.5MHz(2.5-3-2-5-1T). The one thing I think that could be a problem is the HT link being too high @ 1012....my board is only rated at 800MHz. You think that might be the cause of failure. I've noticed though that when I have forgotten to lower the HT multi when OCing, that if it's too high the computer usually won't post. Thanks for listening and contemplating the reason for my failure.
Is it common with socket 939 to OC the cpu to it's limit, AND have the RAM OCed to it's limit? Or do people usually end up with the RAM at stock speed?
EDIT: Right now I'm going to drop the HT multi to 3x and lower the memory to the 150 divider which should definitely be under it's rated speed(yup it's 182MHz) then I'm going to run prime again. Any other suggestions, I'll be here on the other computer.
quick question before testing: do you think it is likely that with a lower cpu OC that I could run the RAM at a higher speed and pass prime? Or is the IMC only capable of handling "x" speed with stability, regardless of the cpu speed?
UPDATE: failed again after 1 hour 21 minutes even after this mornings adjustments.
I think my next step will be to resurrect the Abit AV8 K8T800pro, and see if I have any better results with that chipset. That board will allow me to adjust the chipset voltage as well as other things this board does not. I'm determined to succeed @ a stable 2.9GHz OC.
Edited by ny_driver - 3/8/10 at 7:37am