Inconsistent Memtest86 Results

tecninja

Distinguished
Feb 2, 2008
60
0
18,630
Hello i've been having several issues with a Homebuilt system I made about 6 weeks ago. Today my Drama moves onto RAM.

Some Base Specs.

CPU = C2Q Q9450
MB = ASUS P5E X38 MG613
RAM = 2x 2GB(4GB total) DDR800 Crucial Balistix
Vid Card = Single 8800 GT 512 Card
Nothing is or has been OC'ed ever (yet)


Now I had several other issues, and having thought them resolved, I started running my system with a few new parts this week (PSU being the main swap), I started running into random program failures in Win XP, including games, Antivirus, and Firefox browser. Lots of "program has had a critical Error and must end", and in High end games, often Brief BSOD followed by auto Reboot (like you barely see the BSOD before it reboots itself).

Now I didn't have this problem with the original system setup, but I had power on issues, and eventually It didn't want to power on at all, so after playing with MB and PSU, I decided to RMA the PSU(PC P&C Silencer 610W). Currently I'm using a rather crappy Raidmax 500W that came with the case, but if the specs on the PSU are honest, it should more then run my system, which is probably drawing like 300W tops.

So anyhow with this new random behavior I figured testing RAM made some sense, and I put Memtest86 2.01 on a bootable CD and let it have at it for 14 hours or so. The first pass I got about 50 error hits on test #3, all at 2191.5 MB on test thats supposed to give 20202020 results but the 2020(XX)20, characters were off. Then the test proceeded to run 15 more passes of all 10 tests, with no errors at all.

Memtest86 is reporting the memory as running at 400mhz and 5-5-5-18/ Dual Channel, though my FSB is 333 (not sure if its accurately reporting these values).

I guess my main question is if it is a RAM problem, what are the Odds that its going to run 20+ complete passes with no errors (running more tests now), and What are my Odds that its the MB or **** PSU generating the errors? I'm having a hard time testing if the errors don't come up regularly (IE once in a 24 hour period), and I'd rather not just start Randomly RMA'ing things.

The MB would have to be RMA'ed through ASUS (which seems like an epic pain in the .....), but I think both NewEgg or Crucial will still RMA the RAM. Just wondering if anyone has any thoughts or Ideas.
 

einstein4pres

Distinguished
Oct 11, 2007
311
0
18,780
Your ram can run at a different speed than your FSB. Check your options in BIOS, and make sure that the voltage, timings, and speed are set correctly.
 

tecninja

Distinguished
Feb 2, 2008
60
0
18,630
OK, everything was set to [Auto] in the bios, and as best as I could figure that was 5-5-5-18, 2T, 400 Mhz, and 1.8v

I ran the RAM tests changing some of these values, and found that it seems to run clean at 4-4-4-12, 2T, @ 400mhz at 1.8v (I've only had a few hours to test it so no 24 hour block of "clean" only 2 or 3 complete passes of all 10 memtest86 2.01 tests).

I found it does not like to be overclocked past 400mhz with the other settings held constant, i got it to run at 443mhz (DDR2 887) but I got a few errors in testing, and it wouldn't post at all at 418mhz (DDR 837) at all.

I picked the 4-4-4-12, 2T @ 400 mhz because thats what the stuff was sold as and rated by Crucial. Actually they didn't specify 1T or 2T, and it says 2.2v in the spec. I figured the MB wants 1.8v as its "standard" so i'd start there.

Anyhow if its "Bad" RAM I'd expect to see more consistant errors, then just one test before (at 5-5-5-18) and the ones I got while trying to OC the Mhz. I'm gonna let this current battery of tests run awhile and see what I get.
 

tecninja

Distinguished
Feb 2, 2008
60
0
18,630
OK, I figured to test the RAM, i'd up the settings on the RAM and let it run for a good long period. I set it to 4-4-4-12, 1T @ 400Mhz and manually set it to 1.8v the lowest setting the MB would allow (the RAM specifically says its rated to 2.2v on crucial's site).

With these settings I was able to run for 48 hours, over 70 complete passes of Memtest86 v2.01 without a single error. At this point I would think it is unlikely that bad RAM is what was causing my issue.

Not sure what the source of my problems are at this point.