0

First I had 4x2gb corsair and my system was running flawlessly.

Then I bought myself 4x4GB Corsair 1600Mhz 9-9-9-24 Vengeance DL9 (LP), and when I put all 4 modules in (thus removing all four old memory sticks), linux "free" reports I have 16gb ram, so that's fine, but the whole system froze a few times, and I had to reboot. Didn't find anything in the logs.

So I ran memtest86+ v4.20 with all 4 modules in, and it reported an error, but when I started testing each individual module one by one (3 passes), no errors. Do I need to run more passes or could there be something else wrong? Is it possible that a memory slot is broken? It doesn't make sense because my pc worked with the 4x2gb sticks.

Please advice what to do.

Motherboard: GA-MA785GMT-UD2H

1 Answer 1

0

When both slots of a channel are populated, the tolerances are much tighter and settings that work perfectly with one slot per channel populated won't work with two slots per channel. If you're doing any overclocking or overvolting, undo that and test first with BIOS defaults.

Perhaps you had some specific settings for your previous RAM that you forgot to undo before swapping the RAM sticks. That's pretty common.

Also, it may help to remove the memory sticks, make sure there's no dust or debris in the slots, and then reseat them.

Do not boot an OS you care about until you can run three or four hours of memtest86+ without a single error.

9
  • Interestingly, I sometimes only get errors on my tests about 10-12 hours in... Three or four may not be enough?
    – Bob
    May 7, 2012 at 5:22
  • That's extremely unusual. I wonder if the issue might be heat, noisy power, poor power supply regulation, or something other than the usual issues. May 7, 2012 at 5:23
  • Bob, that's another concern I've noticed. Some people say you should run the test overnight or even 24/48 hours, but other say that such long tests may actually damage the memory.
    – algernon
    May 7, 2012 at 5:25
  • I'm not doing overclocking or overvolting or anything like that. I've been through the BIOS settings, running on default now. I've also vacuumed the memory board so there should be no dust. David, I cannot see why heat would be a problem, bios reports really low temperatures and I have good cooling. Im not sure what you mean by noisy power.
    – algernon
    May 7, 2012 at 5:27
  • @algernon please use @ in front of the name of the person you're speaking to, or we will not be notified of your reply (this does not apply to the owner of the question/answer you're commenting on). Unless there's overheating, I don't see how extended memory tests can damage (good) memory: there are servers (and even desktops) that run 24/7 with constant memory usage. RAM is designed to be used, often.
    – Bob
    May 7, 2012 at 6:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .