Failing Memory, Motherboard or CPU? - The diagnostics so far

[BezzleGear]

Hi,

I wonder if anyone can help me solve a Blue Sceen Of Death problem with my new system. I'll dive right in with the detail.

I have been getting intermitent crashes either during XP load or once loaded (it can crash when doing virtually anything - e.g just closing a browser window) note that I am not overclocking at all (yet). So I ran Memtest86 & Prime95 - both showed errors (Memtest86 fails it on test 6 & Prime95 fails with fatal rounding errors). I have done all the usual as far as getting latest drivers - so I thought I would try swapping out the memory. I was running with OCZ 512MB EL DDR PC-3200 Dual Channel Platinum and swapped it for Corsair Platinum TwinX 512MB DDR XMS3200LL. But running Memtest86 on the Corsair still generates errors (fails on Memtest86 test 7).

I would like some suggestions as to what to try next. My only thought is to swap out the CPU (more money). Are their any diagnostic tools that will help me point the finger at the CPU?



Asus A7N8X-E Delux
AMD Athlon XP 2800+ OEM
Zalman CNPS7000-ALCU Cooler
Corsair Platinum TwinX 512MB DDR XMS3200LL
Seagate Barracuda 7200.7 120GB SATA 8MB Cache
Asus ATI Radeon 9600XT/TVD 128MB
Antec Sonata Case

[Senor Panadero]

Welcome BezzleGear,

Could be a few things. Some things to start, and rule out:

*Check your cpu temperature in the bios Hardware Manager. See how hot it will go sitting there for a couple of minutes. Let us know that final highest temperature.

*Memtest errors are not good. But you may be able to stop them with some settings. Try raising the DDR reference Voltage a bit, up to 2.8 won't hurt anything. Set Memory Frequency to "100%". Check your timings against what is suggested on the manufacturer homepage. They should be showing up the same, but if not they can be set manually. TwinX3200LL is (top to bottom) 6-3-2-2, you can also try 6-3-3-2. Try removing one stick and testing for the errors. If none, remove and try the other. Try the single sticks in slot 2 and 3 (center and furthest from cpu).

[BezzleGear]

Originally Posted by Senor Panadero

Welcome BezzleGear,

Could be a few things. Some things to start, and rule out:

*Check your cpu temperature in the bios Hardware Manager. See how hot it will go sitting there for a couple of minutes. Let us know that final highest temperature.

*Memtest errors are not good. But you may be able to stop them with some settings. Try raising the DDR reference Voltage a bit, up to 2.8 won't hurt anything. Set Memory Frequency to "100%". Check your timings against what is suggested on the manufacturer homepage. They should be showing up the same, but if not they can be set manually. TwinX3200LL is (top to bottom) 6-3-2-2, you can also try 6-3-3-2. Try removing one stick and testing for the errors. If none, remove and try the other. Try the single sticks in slot 2 and 3 (center and furthest from cpu).

Thanks Senor Panadero for pitching in - much appreciated.

The cpu temperature in the bios levels out at 39C after five minutes. The highest I've had the temperature from within Win XP is 49C using Motherboard Monitor 5 - and that was after running Prime95 for several hours.

I was about to try your suggestion but I thought I would run Memtest86 once more so I had a pre/post test benchmark. Frustratingly Memtest86 ran six full test passes without error! I don't know what to make of this as yesterday it would fail on Test 7 within minutes. I think I'll try putting the OCZ memory back in and see if it still fails on Test 6.

What do you recon - was it just fluke that I hit the errors so quickly yesterday? Should I leave Memtest86 running for longer before swapping modules out again?

Several Hours Later:
I put the OCZ back in and guess what - no Memtest86 errors. I hate this kind of problem. So I put the Corsair back in and will run with this and see if I hit the problem again. All I can think is that maybe one of the memory slots is suspect and that repeated adding and removal of memory modules has cleared the problem - does this sound likely? (any way fingers crossed for the moment)

[Senor Panadero]

Originally Posted by BezzleGear

Thanks Senor Panadero for pitching in - much appreciated.

The cpu temperature in the bios levels out at 39C after five minutes. The highest I've had the temperature from within Win XP is 49C using Motherboard Monitor 5 - and that was after running Prime95 for several hours.

I was about to try your suggestion but I thought I would run Memtest86 once more so I had a pre/post test benchmark. Frustratingly Memtest86 ran six full test passes without error! I don't know what to make of this as yesterday it would fail on Test 7 within minutes. I think I'll try putting the OCZ memory back in and see if it still fails on Test 6.

What do you recon - was it just fluke that I hit the errors so quickly yesterday? Should I leave Memtest86 running for longer before swapping modules out again?

Several Hours Later:
I put the OCZ back in and guess what - no Memtest86 errors. I hate this kind of problem. So I put the Corsair back in and will run with this and see if I hit the problem again. All I can think is that maybe one of the memory slots is suspect and that repeated adding and removal of memory modules has cleared the problem - does this sound likely? (any way fingers crossed for the moment)

Yup. Tough to diagnose the intermittent problem.
Test 6 gave some problems in the past with Bartons that I recall. Are you using the Memtest from www.memtest.org? It has been updated, don't know if that will make any difference.

Also, another nice Mushkin article concerning Memtest and errors. got it pasted in the sticky post at A7N8X->Common Questions->"What about memtest Errors?"
http://www.amdforums.com/showthread....76#post1936276

[BezzleGear]

Originally Posted by Senor Panadero

Yup. Tough to diagnose the intermittent problem.
Test 6 gave some problems in the past with Bartons that I recall. Are you using the Memtest from www.memtest.org? It has been updated, don't know if that will make any difference.

Also, another nice Mushkin article concerning Memtest and errors. got it pasted in the sticky post at A7N8X->Common Questions->"What about memtest Errors?"
http://www.amdforums.com/showthread....76#post1936276

Thanks for the Memtest tip (I was using the old Memtest86) and thanks for the link. I'm now beginning to understand why getting your home build pc running smoothly is not just about sticking with the bios defaults! Which in a strange kind of way makes it a lot more fun!

By the way if I get the same problem back again should I start a new post or keep adding to this one?

[banderbe]

Can you make changes to your BIOS and reboot without problems?

I experienced something very similar on my A7N8X-E Deluxe. It started rebooting itself for no reason, so I ran memtest and it found all kinds of errors in brand new Mushkin PC3200 Level2 BH-5 sticks. Then, I tried to lower the memory timings from stock to test again and then I didn't get a post. I cleared the CMOS, and the board posted, but said "CMOS Checksum failure". I could still press f1, boot into Windows, etc., but if I rebooted the board wouldn't post and I'd have to clear the CMOS again. I also couldn't make any changes to the BIOS without having to clear the CMOS again. I called Asus and they are shipping me a BIOS chip free of charge. Also, the memory tested fine if I plugged in just one at a time into DIMM1.

[BezzleGear]

Originally Posted by banderbe

Can you make changes to your BIOS and reboot without problems?

I experienced something very similar on my A7N8X-E Deluxe. It started rebooting itself for no reason, so I ran memtest and it found all kinds of errors in brand new Mushkin PC3200 Level2 BH-5 sticks. Then, I tried to lower the memory timings from stock to test again and then I didn't get a post. I cleared the CMOS, and the board posted, but said "CMOS Checksum failure". I could still press f1, boot into Windows, etc., but if I rebooted the board wouldn't post and I'd have to clear the CMOS again. I also couldn't make any changes to the BIOS without having to clear the CMOS again. I called Asus and they are shipping me a BIOS chip free of charge. Also, the memory tested fine if I plugged in just one at a time into DIMM1.

Thanks for the input - and yes I have been able to change BIOS and reboot.

The situation at the moment is somewhat ambiguous - the calm before the storm I think. I have experienced several boots that don't post - typically three or four at a time - I get to the point where I'm saying what the hell is going on and it posts okay. Then last night I left it running Memtest86+ and in the morning it was just sat their locked up in Memtest86+ after having run for an hour and a half (no reported memory errors though).

Not being able to reproduce the error to order makes it hard to know what to do next without just blindly replacing CPU, Motherboard or Power Supply. At least I am as certain as I can be that its not memory having tried two different dual channel brands (OCZ & Corsair). I guess I'll just have to sit here and wait until something breaks completely!

[BezzleGear]

I've made no more changes and have had no Blue Screens Of Death since the weekend and Memtest86+ runs all night (all tests) with no errors. So where has that intermitent problem gone? - answer - the graphics card. But is it a symptom or the cause? I have experienced a dozen or so boot failures where the grahpics card fan spins up for maybe 5 seconds and then the fan cuts out - all power to rest of system is fine - case and cpu fans running okay, eventually the system boots okay and is stable all day. So presumably not a COP issue. Maybe this failure gets closer to why I was having crashes? What does anyone think?

[molestedbyasus]

1 quick question .. when you say test #6 ... do you mean the 6th test, or the 6th pass?
I usually let memtest go for 6-8hrs, and a few dozen complete passes when I get the intermittent problems, with only bare minimums in (CPU, MEMORY, CD-ROM, Video card... no HDD, no sound etc..).

That way I dang well know whether I'm looking at a problem in memory--->mem controller(northbridge) ---> CPU area, or elsewhere.


But onto the video card ... yes it can cause BSoD's, especially if the fan starts failing, maybe lube it up a little.

That said ... what is your powersupply? A fan failing, and interrmitent memory failures can possibly be low power. I would look to the video card first though.

[BezzleGear]

Originally Posted by molestedbyasus

1 quick question .. when you say test #6 ... do you mean the 6th test, or the 6th pass?
I usually let memtest go for 6-8hrs, and a few dozen complete passes when I get the intermittent problems, with only bare minimums in (CPU, MEMORY, CD-ROM, Video card... no HDD, no sound etc..).

That way I dang well know whether I'm looking at a problem in memory--->mem controller(northbridge) ---> CPU area, or elsewhere.


But onto the video card ... yes it can cause BSoD's, especially if the fan starts failing, maybe lube it up a little.

That said ... what is your powersupply? A fan failing, and interrmitent memory failures can possibly be low power. I would look to the video card first though.

The Memtest errors were happening on test #6 (not pass 6) within a few minutes of starting the test.
The power supply is an Antec TruePower 380 Watt. All components are only 3 months old so I think I'll try and return the video card seeing as I can clearly identify it as failing (even if I don't know exactly why).
Do you have any suggestions of how to go about identifying a suspect power supply?
Thanks for the reply.

[molestedbyasus]

Well, if you'd said its a 300W no-name brand, then I would have started to suspect it might at least be contiributing to your problems. Usually the best way to check is a multi-metre, but...

If you are having random blue screens, and a lot of hardware, reduce the power requirements of the PC. Things like popping in a slower videocard, taking out extra HDD's or anything else you can do without. If your BSoD's disappear after that, then I would be very suspicious about the power supply.

[BezzleGear]

Originally Posted by molestedbyasus

Well, if you'd said its a 300W no-name brand, then I would have started to suspect it might at least be contiributing to your problems. Usually the best way to check is a multi-metre, but...

If you are having random blue screens, and a lot of hardware, reduce the power requirements of the PC. Things like popping in a slower videocard, taking out extra HDD's or anything else you can do without. If your BSoD's disappear after that, then I would be very suspicious about the power supply.

Thanks - top suggestion. I do run two SATA drives, a CD/DVD reader and a DVD writer. I can try running with one of each and see how it goes.