Motherboard damaging memory which then damges other mobos?

Here's the insanity.

Some time ago after installing my Opteron 1212 into my Asus M2N32-SLI (BIOS 903) the PC would randomly power down. I troubleshooted every conceivable problem and switched out memory, graphics, except for the Opteron as I had sold my Windsor AM2.

Anyway recently the board has been having serious memory issues, such as anytime I change the settings the BIOS beebs of death would follow. I would switch memory which would solve it sometimes or take one stick and move it from dual channel mode and run the board off one stick in non-primary slots. That would usually work.

Then I thought it was my memory as one or another stick would seem to create the problem. But when I tested the same kit in my Asus P5W DH (BIOS 1901) there would still be problems. then either one or the other motherboards damaged a stick of Mushkin while the other is fine. Mushkin replaced the kit fast for I'm doing a review but now I'm afraid it is the boards, this is why.

I recently got rid of Crucial here that was working fine in both boards although there was still thge occaisional memory bbeps in the Asus M2N32-SLI boards but it was running fine in the P5W DH before I shipped them out.

The person gets the memory and it won't boot in either of his mobos DFI and won't pass Mem-test. Now he had to RMA one of the baords because it won't won't work properly after having the memory in it once?

I need serious help on trouble-shooting or if anyone has the experience and knowledge to know if this is possible. Two seemingly working mobo's one having some BIOS code problems after memory changes and always in Dual mode are working now, but their seems to be a pattern. I never had to return money before and I can't beleive this has happned. The Crucial Ballistix is en-route and I'll have them Fruiday, but in the interm I'm terrified to use anything.

I have reviews to do so this could kill me. I have a basically unused Gigabyte P965 DS3 so I can salvage that, but will I place a stick in that board which was in the Opteron Asus system and it might damage that board?

PLEASe HELP ME

I had the M2N32 do that once. It just would not post after running for months without a hiccup. However, a CMOS reset and reseating of the RAM solved my issues. No overclocking was done because this system was built for work. The ram was generic Value stuff.

I've noticed that the mem-slots on the ASUS board did not offer a snug fit. That could be a possible problem.

"Snug" you have to lay the case down and jump up and down to get memory in the board, and taking them out is like trying wrassle an elephant out of a Black Hole by it's tiny tail.

Dam things take the gold plating right off the contact pins.

I tried soap and water to get them to go in and out easier (just kidding)

Thanks for replying though.

BY the way I'm running Mem-test on the M2N32 right now and there's been no errors for 4-hours but how long does Mem-test need to run? I've fogotten and does it show errors immediately? If not then at least this stick seems fine. Now I wonder if testing the stick or any stick in one of the Dual Channel slots will show an error IF it's the slot, but not the memory per se? Can anyone tell me about Mem-test's process how long it runs? Here's a photo of the screen as it stands....

lol. If you try applying a lateral force you may break something.

Looks to be a one off problem in your case. I have not heard reports of the M2N killing RAM.

Well I guess there's the possibility that the Mushkin stick simply failed since that screenshot above is of the remaining stick and is running on the M2n32.

There's also the possibility that the memory I sold is OK but wasn't compatable with the person's boards, however; I trust that person and he ran mem-test, although i forgot if he said there were many errors or it wouldn't run. I know he said on one board it began to boot but the point where the memory amount usually takes about 5-sec to caculate took forever.

It's hard to test without proper hardware. Thgis has inspired me to get pro-RAM tester and DIMM tester.

BUT I still need help here people. ny idea how to systematcially trouble shoot this?

At the moment I have a kit of Crucial Ballistix Tracer PC2-5300 running at DDR2-800MHz in my Intel rig and it's fine.

In the M2N32 I have the single stick of Mushkin XP2-8500 running in DIMM 1 of the dual memory DIMMs and no problems, GREAT I thought it was in the single DIMM 3 or 4.

HEY I just discovered somethuing, the Mushkin XP2-8500 heatspreaders use metal (not aluminum) clips, is it possible since the DIMM's on the M2N32-SLI are so close together almost all memory makes contact a short could occur?

BY the way I read your reply wrong, your M2N32 had loose fitting DIMM's? Mine are so tight it's almost impossible to get the memory in. The baord flex's when your trying push the memory in and I've tried every trick in the book (short of soap and water).

Keith,

With respect to your question about Memtest...you need to run something like 32 full loops in order for all of the memory to be tested. I remember reading that in a thread over at BleedinEdge. I'll try to find the link and post it.

Sorry to hear that you lost one of the Mushkins...they were a real nice set.

I can't believe that a memory module could damage the motherboard unless the stick was faulty and created a short...which brings me to your question about whether or not a short could result from the two sticks touching. That is certainly an interesting notion, but one that you could prevent by slipping a non-conductive layer between the modules.

Edit:
By the way, most (if not all) of the ram that I have with heatspreaders has metal clips.

Quote from linked thread (i4memory, not BE):

Test 4 [Moving inv, 32 bit pat, cached]

This is a variation of the moving inversions algorithm that shifts the data pattern left one bit for each successive address. The starting bit position is shifted left for each pass. To use all possible data patterns 32 passes are required. This test is effective in detecting data sensitive errors in "wide" memory chips.

so if you're running standard test #1-7 the bare minimum passes you're want to run would be = 32 passes to test all data patterns

http://i4memory.com/showthread.php?p=6#post6

Thanks I almost got that far the tests was running over 6-hours...

This is certainly a conundrum and it's imperative I rseolve it, especially with my schedule backed up.

I posted this at Xtremesys, and didn't get a single responsre. I used to really like Xtremesys but the last few threads I created there asking for help went un-answered. Here' someone always tries to help.

yea xs is turning into if you dont have new hardware about to comeout or something along those lines. they dont care, imo... i started a thread over there about the $40 D9's i got and my highest oc i tried on the ram before bed. woke up no one even responded/posted about it.

If you don't have phase change tubes cooling your videocard, CPU, NB, SB and one growing out of your Butt they won't even read the post.

I made that my home site back in 2002 and because I chose not to run with the Herd (often over a cliff) perhaps I've offended the wolves.

I always thought the point of these forums was to perpetuate the DIY hobbyist those whom have a passion for something and love to share. Those who come for answers are the most important people here and those with experience remember when they needed help and choose nowto reciprocate.

It won't work any other way because it simply begins to look like other forums where there's more stickies then threads with "Post your fastest speed, greatest bandwidth, coolest temp, fastest Pi, most 3D points." and you find in every one of those posts the same 20-people all staffers defining their manhood by the size of their videocards or futuremark scores.

When I came here it was like a breath of fresh air, as opposed to:

http://www.xtremesystems.org/forums/showpost.php?p=2168162&postcount=3

My next project: how to makje an evaporator bong out of an enema bag (have to use a European waterblock though need 10mm adpatability).

I could make the same complaint about OCF whenever I post something extreme...the audience at XS takes notice of a different genre of results than OCF does, just the way it is, different strokes for different folks. Doesn't mean they're bad people.

Liquid,

All components draw power from the power plane. Depending on the PCB design, you either have a single PP or multiple PP's. Single is most common and efficient. By definition, dirty currents flow through the power-plane. It is highly unlikely that a short in the memory module would initiate a short between GND and the power-plane to take out the whole board.

Also, you have filtering and regulation circutry that has to be bypassed/shorted out. I'd imagine that the MB maker would have planned for such an eventuality and would have included a bleeding path if this occurs (very easy to do).

I am no expert, heck I know almost nothing about the actual art of MB design, so I may be wrong here.

i like seeing what you do Gautam ,but you are right not everyone here is into that kinda thing.

Liquid3D and I have both been around for a long time...him longer than me. At first I'd get frustrated with XS for similar reasons, then lately, with OCF. Now I've realized they really are two completely different audiences. Some of the stuff posted at XS would belong better at a car forum than it would here. Part of the reason for no replies at XS for this thread might simply because practically no one there runs AM2. And the D9NHL's...well, your results are nice but a couple of people posted practically the same speeds with the same IC's for the same price just a little bit ago. Anyways, enough of that.

L3D, what sort of vdimm have you been using? I didn't see you mention it and high vdimm can do a lot of weird things to D9's.

I had somthing similar my Eather somthing happend to my MB or ot my ram but one stick died . now only one out of 4 slots on the mb work =( .

Have you Read about alot of other ppl having the very tight dimm slots on that board ? Maybe there is a flaw with yours and a RMA would fix all your worries .

Gautam said:

I could make the same complaint about OCF whenever I post something extreme...the audience at XS takes notice of a different genre of results than OCF does, just the way it is, different strokes for different folks. Doesn't mean they're bad people.

Click to expand...

I don't believe anyone is "bad" at least I don't know if anyone there is for sure, i do know I've been with XS for over 5-years, and I've seen these changes and in fact I've written about this before over there when they occurred.

I understand things change, and I accpet it, but of course I also have the right to express my feelings about this and knowing others feel that way and may not speak out perhaps I'm validating their conerns, if only that then I've helped another.

Your right about "XS doing their thing" what I'm saying is there was most definately a time when that wasn't strictly their "thing" and as they grew they still were humble enough to help others. That spirit has seemed to left that site and I'm merely expressing my sadness for this. I was joking before and apologized if I "hurt" anyone feelings.

So instead of complaining I'll simply post there more often helping others whose posts go unanswered.

So to answer some questions. I haven't excceded 2.3V there were a few times overclocking the memory wher I tried 2.4V but only for a short time as the memory performed worse and loves low voltage. The entire time I was testing the Mushking, Super Talent and Crucial which seemed to have failed in someone else's system the voltages remaioned at 2.2V and 2.3V but the latter was rare.

Now I'm testing Super Talent DDR2-1000 in DIMMs 1,2 for Dual Channel mode and have had no errors in 5-hours.

I do notice when I'm in the BIOS the system will freeze from time to time. I cannot clearly identiofy any one reason for this. It occurs with different memory and at different speeds. As i said many of these errors seemed to occur after installing the Opteron 1212. Now I always leave the Vcore on Default believing it would remain at 1.320V however I notice CPU-Z show 1.44V and this was validated in the BIOS. This is usually when the CPU ios overclocked either to 300FSB or 250FSB which I thought it was doing at stock because many times it showed 1.320V.

I am beginning to think both the mobo and CPU may have problems, and I don't know if it's more the mobo or the CPU at this point. I also belaive the kit of Crucial I have coming back may have been faulty.

Of course there's also the other varbale which is the P5W DH + E6400 which is rarely overclocked but easily does 420FSb at default Vcore (this time for real). Both systems are water-cooled with externally mounted radiators and temps are always good because I love the cold. Eevemn innthe winter I crack windows for fresh air and right now it's 70F outseide but I have the AC on.

There was a report filed by one of the OCZ guys stating that swapping out DDR2 with the 5VSB active can kill it. The report recommended that the PSU be turned off completely before any swapping was done.

yea i remember that statement it was from Tony with OCZ.

RE-cap

several months ago I get a Opteron 1212 for my M2n32-SLI (BIOS 903). Shortly thereafter i experience random shut downs. While I should have RMA the CPU I had a memory review very overdue and went forward testing Crucial Ballistix PC2-8000 (1000MHz) and Super Talent PC2-8000. I began to discover many instances of "BIOS BEEPS of DEATH" while changing timings and switching out kits. It became rather frustrating to the point I thought a stick of Crucial was faulty.

Then testing on the Intel P5W DH E6400 board seemed to prove this out. I had troubles on that board. I had to take a stick of another memory and insert it into the non-dual channel slots to get the board to recover without clearning CMOS. Since clearnign cmos now involves removing the dam battery on most boards (although I should find out why) it's so tedious I don't do it, after having problems or switching between kits. Also most boards have the automatic recovery feature where you power down the board at the PSU wait a few seconbds and it res-sets to default mode although your prior settings remain.

The review was finished and given the number of times I switched memory and number of times I had to change timings only to get a BSOD and then re-set againn I realized many of the "errors" I was getting may have been normal given I was testing both kits to their limits. The review was based on maximum timings at several speeds but trying to keep the memory at the manufacturers specified 2.2V. So there was no overolting which would have damged anything.

OK then I send Hardass a kit of Crucial Ballistix which was running perfectly fine in both boards (once a setting waas chosen) and prior to shiipping it out (as I always do) I ran Orthos on the kit on my E6400/ Asus P5W DH (1901) combo, this was his reply;

> I installed the ram in my DFI Infinity 975. It will not run. I get long
> beeps. tried several times.. Take it out put OCZ back in no problem. I
> installed it in my DFI 965-S it takes forever for the memory test during
> bios screen. it fails memtest within a minute of start. This is at 40ox9
> with ram set to 533. Looks like we have a problem

In a phone call he had stated he had tried the memory in a board which then wouldn't boot at default settings? he actually RMA'd the board for which I offered to reimburse him for shipping. While I'm concenbred about my rigs since I use them for my reviewing, I feel horrible which is what prompted this post. I've sent Hardasfunds and offered to cover any expenses he's incurred if the memory I sent him damaged his board. This is my fear, somehow a damaged motherboard is in turn damaging memory which has intrun damaged someone's mobo? Does anyone think this is possible? this is why I'm spedning days running mem-test but at this point I wonder if I can eliminate the M2N32SLI? It hasn't shut down randomly, and I wonder if it was shutting down randomly because it had the Crucial in it? Perhaps the Crucial Kit is dmaged that would explain Hardass's problems, and my problems. Especially with opetron as an on-die memory controller reading errors would shut down the system.


reason I freaked is because I had the feeling the baord may have damging memory, my reasoning for this was the shut-downs I mentioned, instances of memory error codes, and a stick of Mushkin XP2-8500 which fried after it had just ran at 1200MHz at 2.30V in the M2N32-SLI baord. A new stick was sent.

Then I tested the remaining stick in the primary Dual Channel slot of the M2N32-SLi and it ran 16-passes error free. now I just finished testing a kit of Super Talent used in the review on the boards mentioned 26-passes error free.

So if it was the mobo wouldn;'t there be immediate errors on the memory of the DIMM's were damaging memory? I'll have to test the Crucial when it gets back, but could it be possible not clearning CMOS by Hardass contributed to his errors? I mean in most cases I find mobo's recover on their own. If something is wrong usually you power down and when you re-start the aumatic recover kicks in placing BIOS settings to a default state. Where the memory should have ran. here's the last Mem-test;

Did you even read any of the other posts/suggestions?