Q6600 failureclocking

[MURDEROUS G]

Just put together a Gigabyte P35-DS3P, Q6600 and 4GB OCZ DDR2 800MHz. Trying some fairly tame overclocks at the moment, and I don't seem to be able to get even a mild OC to remain stable. At 2.8GHz (400x7) with 1.4V vcore, the system falls over and dies within a few seconds of starting a Prime95 'In-place Large FFTs' or 'Blend' test. The RAM is unlinked and set to 2.0x, which should be OK, right? (I am relatively new to overclocking so apologies if I have neglected to mention something obvious)

After doing some searching for posts by other people who've had stability issues at such relatively low speeds, and reading this thread, I tried running a Prime95 'Small FFTs' test and that appears to be perfectly stable. What does this indicate?

Thanks in advance,
mg

 

[MURDEROUS G]

Hi mk17,

Thanks for your reply.

run small-ffts, with "Round-off Checking" on the Advanced menu, and make sure you've set ram volts (vDimm) to c.2.10v. 2.8GHz should be a given out-of-the-box for Q6600.

The DS3P has a 'DDR2 Overvoltage' option in the BIOS. Setting this to +0.3V should give vDimm 2.1V, right?

If you have 2*2GB ram, put them in alternating sockets, not neighbouring ones. if you have 4*1GB it doesn't matter.

Yep, that's how they're configured currently.

Ideally run a full memtest at stock settings to make sure you're starting from a stable position. At intervals during your o/c hit small-fft's (or orthos) to ensure stability.

I downloaded memtest86+, burned the ISO and am running it on the first of my two 2GB sticks now. Less than halfway through the first pass, it has encountered 83 errors. I guess we have identified the problem?

 

[MURDEROUS G]

If you ran the memtest with default memory timings / volts and encountered this, then this is the problem. You need to set up an RMA to get the memory shipped back to the supplier.

I wouldnt bother slacking off the memory timings to get the sticks to work, if they don't work at the stated speeds then get them replaced.

I figured that too, but then I took a 1GB stick from an older machine, which I know is good, and memtested that too just as a control. Again, hundreds of errors per pass, and all the same error as before. I googled the error, and it seems to be a known false-positive issue with certain versions of the DS3 BIOS.

Anyway, long story short, I used Gigabyte's own Windows BIOS update utility to flash the BIOS to the most recent version to eliminate the false positives, and the BIOS update promptly failed. So now I have something rather more immediate to worry about than the RAM.