It's all on a test bench, and the only thing that will change in the final iteration will be the PSU, which will be my Silverstone STF500, and my case, a CM690.
Everything boots up fine when I have ONE stick of RAM in (either one work NP), but when I put two sticks in, the system "boots," but the CPU fan doesn't start, and nothing comes on screen.
My first thought was to ask Mushkin Greg at NCIX forums, who graciously provided me with correct timings for stuff that I neither know about nor understand, but can follow directions to set (such as ACT to ACT, Rank write to read, etc). I change those to his specified values, and the system won't boot right up with ONE stick of RAM, let alone TWO; it boots, reboots, and reboots again immediately, which I have learned is " you, I'm resetting all my memory settings to 'auto'" in Giga-speak.
Has anyone run into this before? Can anyone hear my head hitting the wall? How bad is it when you can't see after banging your head against the wall for a while?
Oh, sorry. I should really have put that in in the first place, shouldn't I? I've got the memory at 5-4-4-12 @ 2.1v (though I can't directly set this, I have to set an "overvolt" of 0.2v) The settings it keeps reverting to are 5-5-5-18, but it won't boot two sticks with that setting either, regardless of whether or not I set an "overvolt."
I meant all voltage settings. The ram sticks may be at fault here, but I'm not convinced. It could also be either a BIOS voltage setting that needs adjustment, or a failing PSU.
The sticks are reverting to 5-5-5-18 because those are the timings automatically set by the BIOS from the sticks' SPD.
Oh. OK. My vcore is 1.204v, my DDR2/3 claims to be running at 2.112v, my 3.3 is 3.312, and my 12v line is 12.365, which is a TOUCH high, but still well under 5% off. There is not listing for my 5v line, for some reason. I also have MCH overvoltage set to +0.2v, which I assume is memory channel, though I'm not familiar with Gigabyte BIOSes (I've always had Asus before now).
Put your vCore on Auto for the time being; vRam is fine; vMch is your NB voltage: put it at +.1V. With everything else on Auto, it should boot. Now if it doesn't, run Memtest86 on a single Ram stick at a time. If there is a single error, the Ram is the culprit.
Once you recieve that STF500, install it and try to boot. This will test the PSU.
Thanks, 3.0Charlie. The vcore is on Auto right now; I'll drop the MCH to .1 and set everything to auto, but I'm pretty sure I've already tried booting like that with no success. I guess I'll have to let give it a real swing at memtest--I only ran it for about 1/2 hr per stick last go.
I've actually GOT the STF500--it's in my main system, while this is on my testbed (I didn't want to tear apart my main comp without being sure the new one would work--a good idea, as it turns out). I'll try to Frankenstein it up and try it.
Last edited by vulcan500rider; November 13, 2008 at 08:17 PM.
An update: both sticks of RAM are memtest stable for 8+ hours on their own, but still will not boot together. This has led me to wonder if it's possible for one of the channels of the memory controller to not work, and the only way I can think of to test that is to start with one stick in the second channel--which doesn't boot. This begs the question: should any of the RAM slots, other than the first, be able to boot the system, or do you always have to have a stick in the first slot? The manual covers dual channel, but says little to nothing about running in single channel (I'll check again in the meantime, however).