Mobo or ram problem? (or neither?)

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
Running Windows 7/64 on a home brew system built early 2007.
Asus P5B-E with latest BIOS
Intel e6600
2x1Gb sticks of Corsair cm2x1024-6400c4
2x2Gb sticks of AData DQVE1B16
5-5-5-18
Corsair 520HX psu
Intel x25m 80Gb SSD
Various sata hd's

I've tried to keep all drivers updated. Computer has run fine but had a recent bsod of the IRQL not less or equal variety. Updated a few drivers and all has been well except for some odd 15-25 second irregular hangs of Firefox and occasionally Chrome which may have developed after updates to Firefox 6.

I ran the Windows 7 memory diag and it reported hardware (not specific) problems. I d/l memtest86+ v4.2, burned it to a CD and began running tests. Thus far i've found all 4 sticks to fail individually in the A dimm slots 1 and 2 (yellow and black). All fail on test 6 at various times, one stick not failing until passes 4 and 6.

I'm currently testing the B dimm slots 3 and 4 (yellow and black). I just saw a fail on pass 3 with test 7 on AData stick 1 in slot 3 as I type this.

How can I determine if I have memory or mobo problems, based on what I'm seeing? The test results seem to indicate mobo failure. Or do they?

I'm not sure there's any point to continuing memtest86 any further.

Any suggestions?

The above was typed on my iPad. Frustrated with being unable to determine a specific problem, I plugged in my ram and cranked the main PC back up and am using it to post now.

I don't know how I can run anything with the errors showing up in memtest86. And yet here I am.
 
Last edited:
The best way to figure out if it's a memory or mobo issue is to get an extra either motherboard or memory. Should be able to pick up something here or on ebay for fairly cheap.

It is unlikely that all four RAM sticks would go bad at the same time, which points to the motherboard.

You can also do the cardboard test. This is taking out the motherboard, putting it on the floor on top of a piece of cardboard (or desk), plugging in only the essentials (power, cpu, memory, video card if neccesary), then turning it back on. If no issues, add one component at a time. I would start by having just one stick of memory at a time. You start with the barebones and add one thing at a time.

If it works on the cardboard, but not in the case, then it's possible you have a short. Something is touching the motherboard to the case in a place where it's not supposed to connect.

Also try changing the power supply and power cord as well.
 
from my experience irql has always been ram related..id get the ram rma'd or buy another set of ram...
cant ahve any failures with ram ..any error is a problem...sometimes voltage too low also..ck in bios and set manually..
its possible it is a motherboard memory controller failing but changing out ram is pretty fast to do...above post is also good advise too
 
Last edited:
Ouch. I agree that it's a very remote possibility that all 4 sticks of ram would go bad at one time. And the cardboard test is something I'd rather not do since the computer seems to be running fine.

I would likely have never had this angst if I hadn't run that memory diagnostic test. ;)

I hate to put any money into DDR2 ram since if I have to replace the mobo, I'll likely have to get DDR3 ram.

I am not looking forward to another system reset. I had two OS hd failures last year (which is why I got the SSD) and it's a PITA to reinstall all my programs and tweak them the way I like. I have my data backed up, but all my custom settings take a while to sort out.

Dilemma.
 
Well, another thing to consider is DDR3 is MUCH cheaper than DDR2 right now. I've seen 8GB of DDR3 for $30, whereas it would probably be at least twice that much for 8GB of DDR2.

My intuition tells me it's probably the motherboard, but no way to tell unless you have another board you can test on. Or at least some different sticks that you know work fine.
 
That sounds like a prod towards a new mobo and ram. :D
Edit: And from what I've seen, DDR2 is about 4x the price if you can get 8 Gb of DDR3 for $30.

But how in the heck can this system run as well as it does with the errors I saw?

Check this out:
DimmA, slot 0, stick 1 and stick 2 of Adata 2GB:
hlkrI.jpg

2UZU3.jpg


DimmA, slot1, stick 1 & 2:
OjDrD.jpg

WfSMx.jpg


DimmB, ,slot2, stick1:
jNOLa.jpg
 
I had exactly the same issue with some OCZ DDR2 ram not too long ago. It would post all these failures in Memtest, but it would run fairly stable in the OS. It would still BSOD every once in a while, but for the most part, it would be okay.
 
Wow. I wouldn't expect it to be stable at all. And yet I haven't had a bsod for quite a while now. And that was a rare event.
 
Update. So much for my stable system. Just had a bsod of the system service exception type.

And I just started a rebuild on my laptop just before the main PC decided to misbehave.

Anybody want some hot stock tips while I'm on a roll?
 
Did the memory previously pass MemTest86+ or MemTest86 and only recently started giving errors? Because that could mean a problem with the power supply (measure with a voltage meter) or motherboard, but I think that particular model mobo should be OK because it's made only with polymer and Panasonic capacitors. I'm not saying the mobo doesn't have bad caps, and I've had other brands of caps fail around the DIMM sockets. OTOH those modules are made with no-name chips (both the A-Data and Corsair), and anything running with good hardware and fails MemTest86+ is pretty bad because I've found that MemTest86+ passes a lot of memory that MemTest86 or Gold Memory rejects. I'd really want to test that memory in another mobo of the same model or at least of the same Intel 965 chipset.

Cardboard, floors, and table tops are generally not anti-static. At the very least, place the mobo on top of the sheet of anti-static foam it was packed with from the factory. Also do not place the mobo on top of any metallic anti-static bag because most are anti-static only on the inside ("metal in"), unless labelled "metal out". OTOH it's OK to turn such a bag inside-out and place the mobo on top of that, provided it's not one of those stiff aluminum foil bags, which will conduct electricity and cause shorts if its fragile plastic layer is pierced. Transparent bags with black lines all over them or pink bags are also OK on both sides. If you don't have any of those materials, put the mobo on top of a 1/2" or thicker layer of newspaper or brown paper; avoid any paper that's shiney on either side.
 
In all honesty, I don't recall whether or not I tested the ram. Remember, this rig was built 4.5 years ago. And just to clarify, I ran memtest86+ v4.20, not memtest86 if that matters.

I suspect the chances of finding a p965 chipset mobo to test with are rather slim. I can check a local shop and see what they'll charge me to test the ram.
 
Don't go to a local shop. Go to the "buying/selling" forum. I'm sure someone has a cheap motherboard that will work.
 
In all honesty, I don't recall whether or not I tested the ram. Remember, this rig was built 4.5 years ago.
Can't you simply look in the detailed leatherbound log you've kept on that computer? :D

4.5 year ago was about the time when 2/3 of the Kingson PC3200 modules I tested would fail. :(

I wouldn't take the memory to a local shop because their testing won't be any more thorough than yours, unless they own a PhD RST testing card.

It's not absolutely necessary to test the memory in another 965 mobo, but certain memory modules work perfectly with some types of mobos but fail with others, although in my limited experience Intel chipsets have been more tolerant than other chipsets.
 
Update: here are the results of my PSU (Corsair HX520w) testing.

To test the PSU, I pulled the ram out and disconnected the power to the 3 HDs in the box. Had to listen to that infernal beeping error code during the time I had the system powered up for the test.

+12v lines measured 12.11v
-12v lines measured 11.98v
5v lines were all 5.01v
3.3 volt lines ranged between 3.390-3.394v
5poHq.png


All the capacitors looked good.. no bulging tops or signs of damage.
s5NYi.jpg

ZAasI.jpg


And after I put everything back together, hooked up power and turned the system on, I got a msg saying the OC had failed. So I went into the BIOS and under AI tuning, set everything to default and saved it. I'll go back later and check the BIOS settings again. But I can't really tell any difference in system performance before or after.. just doing the 'net thing.

No smokin' gun found yet.
 
I just remembered that I have another DDR2 computer with 2 Gigs of ram. I'm testing it now to see how it fares. If it checks good, I'll yank that ram and try it in my problematic PC. And I'll try my ram in that other PC.

It was built about 11 years ago and everything is 11 years old except the mobo, ram and video card. :eek:

Edit: no errors reported after 3+ hours of memtest86+. If it checks good after 7 passes, I'll yank the ram and put it in the troublesome box and see what happens. If it tests bad then, I can think of no other answer but that my mobo is failing. I believe the psu tests good, based on my measurements above.
 
Last edited:
Update: swapped memory around with the other computer. It has an ASRock 4coredual-vsta mobo and won't boot with anything greater than 1Gb ram per slot, so I can't test my 2Gb sticks.

IAC, here's the kicker: the DDR2-333 ram from the spare PC runs without error (9 runs or so) in my main PC. and the 2x1 Gb DDR2-800 Corsair ram from my main PC runs without error in the old PC.

What exactly am I to conclude from that?
 
I hate to put any money into DDR2 ram

Yeah, I wouldn't put any more money into that either.

If it checks good after 7 passes

I know this sucks, but for each test let it run overnight. Record your results.

2x1Gb sticks of Corsair cm2x1024-6400c4
2x2Gb sticks of AData DQVE1B16

Running mixed memory usually doesn't end well. Pick a set and try to get that running with no overclock, bios defaults, and no errors overnight.

You might want to start with the set that will run in both systems.
 
I guess I should add, if it throws errors, no real need to let it keep running.
 
I'm still battling this around. Set my P5B-E back to basics, as far as I could and still keep AHCI set properly, etc. But I put all the chipset stuff in auto. After several hours and 5+ passes, the 1x2Gb Corsair had no errors. I tried the AData 2x2Gb and after a pass or two, got an error. Even though the Corsair is labeled as 4-4-4-12 timings, I couldn't get it to run error free that way. But when I reset the NB timings back to auto, I had no errors after 5+ passes... and yet the cheapo DDR2-333 1x2Gb ram from the other computer ran fine without resorting back to default settings.

My cheapest and quickest option to get this sorted out (assuming it really is a ram problem) is to buy some more DDR2 ram. I'm convinced my psu is good and *think* my mobo is ok. But it puzzles me as to why I'm just now having issues if the mobo timing has been out of whack because it's been like that for a long time and through two OSes (Vista/64-one install, and Win7/64-two installs).

But if it turns out that new ram still doesn't fix it, I will have blown $$ for nothing. It will cost me at least $50 for 2x2Gb DDR2800 ram which I will have no use for if it doesn't work because my other DDR2 box won't accept more than 1x2Gb ram. Anything I buy now will have DDR3 ram.

Can I mix 1x2Gb Corsair with 2x2Gb Corsair? Or do I need to get 4 sticks of ram all the same? (think my existing 1x2Gb Corsair is likely good but that my 2x2Gb AData is bad.)

When I first put it all back together this morning, it wouldn't boot up and wanted to start loading the repair files... where it seemed to hang. So I pulled all the ram out... made the 'BIOS adjustments listed above, and reran memtest86+ on the Corsair (without error) and then on the AData (with errors.) Right now all the ram is back in the box and all is running fine at the moment.

Argh. What to do....
 
Back
Top