Random re-booting/Memtest errors on newly built computer

[chriskder]

My computer was restarting randomly and unexpectedly so I disabled the auto-restart function to see what the error was. When it crashed again, I got a blue screen of death saying that I encountered an IRQL_NOT_LESS_OR_EQUAL error.

I decided to run memtest to see what type of errors I would get. With my memory sticks in DIMM 1 and 3, I got 192 errors in Test 5 and 1 error in Test 8. With the memory sticks in DIMM 1 and 2, I got 2 errors in Test 3, 18 in Test 4, and 1 in Test 6. However, memtest recorded no errors when each stick is tested independently on DIMM 1. How is it that no errors are recorded when there is only one stick mounted on the mobo, but errors occur when both sticks are mounted on the mobo?

===========================

I recently built this computer with the following specs:

AMD Athlon 64 3700+ San Diego 2.2GHz

DFI LANPARTY UT nF4 SLI-D Socket 939 NVIDIA nForce4 SLI ATX AMD Motherboard

Kingston ValueRAM 1GB (2 x 512MB) 184-Pin DDR SDRAM DDR 400 (PC 3200) Dual Channel Kit System Memory

NEC 18X DVD±R DVD Burner With 12X DVD-RAM

Creative Sound Blaster SB0610VP 7.1 Channels PCI Interface Audigy 4 SE Sound Card

BIOSTAR V6202EL63 GeForce 6200 LE 256MB TurboCache(64M VRAM on board) GDDR2 PCI Express x16 Low Profile Video Card

Western Digital Caviar SE WD1600JB 160GB 7200 RPM IDE Ultra ATA100 Hard Drive

Rosewill RP550-2 ATX 2.01 550W Power Supply


Thanks in advance for any suggestions or help provided.

 

[meltdown75]

The last system I had which exhibited this behaviour had a couple blown caps on the motherboard. Check the caps on the board to be sure they are all intact. If any of them appear to have bulged out or are leaky, then you definitely want to get that board replaced.

Were all the components new or did you buy some of them used?

I also experienced random reboots with that system though, something you hadn't mentioned. You might as well see if you can swap the ram sticks you have for others as well. I wouldn't rule out getting an entirely different board if that didn't remedy your problem but I'm skeptical of certain boards so that's just me. Best of luck and post back no matter what.

Cheers

Melty

 

[chriskder]

Originally posted by: meltdown75
The last system I had which exhibited this behaviour had a couple blown caps on the motherboard. Check the caps on the board to be sure they are all intact. If any of them appear to have bulged out or are leaky, then you definitely want to get that board replaced.

Were all the components new or did you buy some of them used?

I also experienced random reboots with that system though, something you hadn't mentioned. You might as well see if you can swap the ram sticks you have for others as well. I wouldn't rule out getting an entirely different board if that didn't remedy your problem but I'm skeptical of certain boards so that's just me. Best of luck and post back no matter what.

Cheers

Melty

I'm gonna check up on the mobo for cap problems. All of the parts are new or OEM, and were ordered from New Egg.

 

[John]

If you're using a new mainboard you will not have blown caps.

1) Try the orange slots 2 & 4
2) Try SPD for the memory timings, or try setting them manually at 3-4-4-8
3) Try bumping the memory voltage to 2.7 - 2.8v
4) Try CMD @ 2T

 

[chriskder]

Originally posted by: John
If you're using a new mainboard you will not have blown caps.

1) Try the orange slots 2 & 4
2) Try SPD for the memory timings, or try setting them manually at 3-4-4-8
3) Try bumping the memory voltage to 2.7 - 2.8v
4) Try CMD @ 2T

Yeah, I just checked and the caps look healthy. In re: your questions:

1) 2 and 4 won't work. The CPU, PSU, etc, everything turns on, but nothing shows up on the monitor.

2) Where in the BIOS do I change this?

3) Where in the BIOS do I change this?

4) I don't get what you are trying to point out here

Sorry, I'm not that technically savvy with building computers. My first attempt (the one I am typing on right now) went smoothly without any hitches.

 

[RebateMonger]

Originally posted by: chriskder
How is it that no errors are recorded when there is only one stick mounted on the mobo, but errors occur when both sticks are mounted on the mobo?

I've seen this before. It's not uncommon. Particularly when Dual Channel Mode is enabled. And it makes it REALLY hard to know for sure which memory module is "defective".

 

[chriskder]

I tried boosting the memory voltage from 2.6 to 2.7, hoping to make the system more stable, but that has not worked. In fact, memtest gave me over 500 errors in test #5 with the voltage now at 2.7V.

Oh this is just getting frustrating.

 

[Gustavus]

chriskder,

From your description, the logical thing to suspect is the memory. Value RAM is often sold at pretty near the spec limits. Ordinarily there are two things you can do to push it a little -- raise the voltage and lower the latency settings. Well, there is a third -- underclock by lowering the FSB. Of course the first thing to try would be to substitute other memory (if you have it or can get it) since that would pin down the problem to the memory. You have already tried raising the voltage, so about the only thing remaining is to raise the latency settings in the BIOS. I assume you are operating at the default SPD speedings now, so set these manually in the BIOS setup. If that doesn't work and you have a CompUSA near you, they can test the memory modules -- and have often done so for me for free. That would make it easier to RMA them since you could then say they had tested bad.

There is one other thing, sort of like grasping at straws -- leave the memory set at the higher voltage and let Memtest run over night. Sometimes when memory is right on the edge, running at a slightly elevated voltage for a time seems to help the problem. Run in or bake in I guess, but I don't know the physics by which a junction would "run in". It does work occasionally though.