memory errors

[AUTOgod]

hi guys.

around a year ago i moved from my first computer ever (got it about 6 years ago), to my second, which was a major change since this one is built by me.

i was ecstatic, and got a q6600, a gigabyte ga-p35-ds3, an 8800gt 512, an antec 900, 4GB OCZ 800 ram, and a hiper 580w psu.

soon after buying these components, memory errors started showing up, so i returned the memory for some slightly different memory (nvidia SLi OCZ 800 4GB). this started to get errors also, so i returned that too, and got 4GB of OCZ HPC reaper 800. just recently, soon after moving into a cosmos 1000, every OS that i am running (vista home premium and win 7 beta), has had the "host process for windows has closed", and the past couple of days, half life 2 and garrys mod have been crashing repeatedly, and randomly. just yesterday x3:reunion crashed while i was playing, so i dropped dropped to windows, only to be greeted with "a plug and play device has terminated unexpectedly" (or something along those lines), and windows restarted. however, during the boot, it blue-screened with "memory management" as the cause. i have checked for viruses several times and have come up clean, and i have reinstalled windows cleanly.

i was running a very stable overclock at 3.00GHZ with the memory at 800MHZ/1.8v (rated at 800MHZ/2.1v) and now i only have slight stability at standard (2.4GHZ/800MHZ/1.8v) and the memory is running much hotter than before (x3 crashed at standard clocks).

so, after that long winded explanation, my question is this:
what are the chances that it is my motherboard at fault?

if so, which would be a good replacement?

i am restricted as to where i can buy a replacement as i need this comp working properly soon as i have some work to do, so the only places i can buy are here:
http://www.anglianinternet.co.uk/

here: http://www.mgcomputersnorwich.com/p...Online_Catalogue_Components__7.html&NOLOGIN=1

or here: http://www.pcworld.co.uk/

thanks in advance for any help

 

[Mussels]

Hiper PSU's have a poor reputation. have you looked into the possibility that your PSU is causing the problems?

Make sure you're running stock clocks and that your ram has its voltage manually set in the BIOS, to whatever it requires.

 

[AUTOgod]

i have set everything to completely standard, as it was before and its still crashing.

i ran a memtest and one of the sticks came back as very bad. also, when i put all the sticks back in to go back to 4GB, it gets many times worse and memtest wont finish at all, and just crashes.

i havent had any trouble before, and it has been rock solid stable before. and since these are all memory errors, i dont think the psu is to blame. if it was, i would get more than memory errors. also, if the psu couldnt cope, then prime95 would crash it right? im pretty sure its a memory problem, but 3 different sets of memory? i think the motherboard. just need some opinions.

 

[Mussels]

define "standard"

Memory voltages need to be set manually. Most of the ram kits you mentioned require 2.1V, default for DDR2 is 1.8v

If your PSU was undervolting the slot, it makes complete sense to me that all 3 kits would have problems? PSU's have 3.3, 5 and 12V rails. Its possible only the 3.3V rail is faulty, for example.

 

[AUTOgod]

the first ram kit was rated at 1.8 and still had problems.

the second kit was 1.9 i think, and this kit was 2.1.

but it has been working for several months set at 1.8 perfectly.

by standard i mean as it came, 2.4GHZ on the q6600, 266 fsb, 800(400) on the ram with 5-5-5-15 timings.

 

[Mussels]

"its worked until now perfectly"
Yes, that IS the definition of "broken"

You need to troubleshoot here. You've tried other sticks of ram, which only rules out those sticks of ram at your current BIOS settings.

Once you've tried higher voltages, you rule that out of the equation as well, making it far less likely to be the memory modules themselves.

With that taken care of, you then start looking at other culprits, such as northbridge overheating, PSU providing insufficient power, and so on.

Ignoring any possibility is a quick path to never getting the problem solved.

 

[AUTOgod]

"its worked until now perfectly"
Yes, that IS the definition of "broken"

You need to troubleshoot here. You've tried other sticks of ram, which only rules out those sticks of ram at your current BIOS settings.

With that taken care of, you then start looking at other culprits, such as northbridge overheating, PSU providing insufficient power, and so on.

Ignoring any possibility is a quick path to never getting the problem solved.

as said before, one of the sticks failed with 2 errors on test 5 of memtest86.

however, with that stick removed (and another, requires matching pairs) it is still intermittent.

checked the voltages in the bios, they read:
12v = 11.985v
3.3v = 3.360v
ddr2 = 1.902v
vcore = 1.188v

added 0.1v to the ddr voltage, and it still reads as 1.902v.

im going to leave it running while i go check if the warranty is still valid (probably not, but hey, worth a shot).

also, you said: "it makes complete sense to me that all 3 kits would have problems"

that would mean that they would malfunction immediately, when they actually go wrong some time after purchase/exchange.

also, i find it difficult to believe that the psu is to blame, as it has been left for 6 hours running prime95 with "in place large FFTs" with no errors at all. it is only when memory intensive applications are used.

 

[Mussels]

that would mean that they would malfunction immediately, when they actually go wrong some time after purchase/exchange.

No, it doesnt mean that.
If you're getting erratic power spikes, different ram modules will take different lengths of time to fail. You're using logic to diagnose this, but you're making lots of assumptions on the way.