This memory is killing my OC!

[Xris]

I have 4gb of mushkin pc2-6400. I got my Q6600 up to 3.2 then above that I got BSODs like mad. I took a stick out and got up to 3.6 then RMA'd the bad stick. So while I'm waiting I run P95 to make sure it's stable for a second time, after having already ran 8 hours once before with no problems. It wasn't stable. Then I start getting BSOD's again, not often but the thing isn't stable at all.

The new stick arrives I pop it in and my win experience score goes from 5.2 to 5.9 (both times I had 4gb's, mem being the lowest factor on the test) I can't OC above 3.0 with these sticks in now.

I had ordered a new radiator and some sleeving so I messed with that for a few days. Now I'm ready to tackle the OC problems again. I bump it back up to 3.6 and get BSOD's again. I took out one stick, the one that I did not RMA and it seems stable so far. I'm about an hour into P95. I don't know yet if the stick I didn't RMA ended up frying while the other was out or if my mobo doesn't want to see two sticks at the same time. I'm leaning toward the stick frying while the other was out.

I've, of course, got some more troubleshooting to do. But what do you think? I've run memtest86 with both sticks in and got no errors at 2.4 ghz. Next step seems to switch sticks out. If I get errors then the problem is obvious: the stick that wasn't RMA'd is bad. But it was fine before. It failed while the other is out. Why? If switching sticks produces no errors then my mobo just doesn't like 2 sticks. How can I fix this?

[JamesXP]

Raise your northbridge voltage first (MCP voltage I think..)

[Xris]

Okay, so I got my Q6600 up to 3.71 with one stick in and ran P95 for 45 minutes without incident. I switched the sticks and hit the okay button in P95 and 15 seconds later it's BSOD time. I'm running memtest86+ with the offending stick right now. How long should I run this? I already ran it at 2.4ghz with both sticks in and got no errors.

Does it sound like this one stick is bad to you all?

[maxfly]

you may need to loosen your trfc.use memset to see what the chipset is defaulting it to.

[Xris]

Oh maxfly, I hope you're right.

[Xris]

At stock speed tRFC was 51
OC'd it was 50 (RAM is at 793?)

I bumped it to 55 and put both sticks back in. I'm running the P95 blend test right now.

[Xris]

It's still going after 10 minutes. It would've failed by now. Maxfly, I think I might love you! I'm going to leave it overnight.

[Xris]

Maxfly, you magnificent *******, it's stable at 3.7Ghz thanks to you.

[maxfly]

:) glad to have helped!

[Blazing fire]

When ocing, I'll loosen timings (5-5-5-15 etc.) and top out CPU before tweaking the RAM. Anyone agrees with me?

Btw, I'll rather everast over vista performance score and memtest86+ over prime blend.

[DocClock aka MadClocker]

With marginal ram, I have found that Memtest86+ won't cause errors till you run the "extended test" or "advanced Test" can't remember which..anyway, just hit the "t" key when memtest starts, and it will run the extended tests, and you will find out in short order weather your ram is bunk.

[Blazing fire]

That's something I'll remember and recommend after I've tested it :). THanks for that input!