I Ran MemTest86 and received errors, now what?

[OverVolt]

Well what do i do!!!

I got 384 errors on Test#4 right off the bat. But then i ran the test again and receive 0 errors for the next few hours. So is my memory bad or not? I've been having probs getting WinXP to boot on this rig (what a wonderful OS it is indeed) No hardware problems just bad SATA RAID drivrers that it doesn't like. So i decided to run MemTest86 for fun (i got bored) and voila, errors.

So what does it mean short of calling up Kingston, and yelling at them for a RMA. Is that what i should do? The computer had stability problems before with reboots crashes ect ect ya know what an instable system is.

 

[hjo3]

How many passes are you running? If you get more than 1 error in 10 passes, it's a good idea to RMA the memory. If you want to be thorough, take the RAM and put it in another mobo and MemTest86 it on c-1-2-2-3-0 for 10 passes. Odds are that it's really the memory though.

 

[Slammy1]

Are you running standard or o/c? PAT enabled? You can try bumping VDIMM slightly (but keep it below 0.3V over spec). I run HX3500 and find I need to bump the voltage to meet chip specs (It tops out around 456DDR, so I guess I can't complain). No matter what I'd do, my system's unstable at default BIOS.

 

[OverVolt]

Well 384 is alot of errors ! there running 2.65v, 2-3-3-7-1T @400Mhz and their DDR434 modules @ 2.65v and 2-2-2-5 so there way under spec. So far it's been 11pases and no errors so far (besides the 384 in 10seconds) could it have been a fluke?

 

[egale]

You can bump the voltage to 2.75 and with absolutely no problem. If you get errors, (or even not), I would increase the vdimm up to that.

 

[hjo3]

> So far it's been 11pases and no errors so far (besides the 384 in 10seconds) could it have been a fluke?

Weird. I agree that it must have been a fluke. Just chalk it up to invisible computer gnomes, I guess.