Please help w/ Memtest86 result

[tja111]

Using v3.2 of Memtest86. Let it run all night = 8 full passes. Zero errors reported.

Changed the test parameters under memory allocation to "BIOS-All". 47K errors reported thru just the first 3 tests.

Memtest recognized my configuration perfectly.

Actuall problem is system crashing during heavy processing. System simply shuts down and reboots. No blue screen, only a cryptic report XP report upon restart. Ruled out SW issues

Opened the box and felt around after crashes for heat. One of my memory modules is too hot to touch. Assuming the failure is heat related. Question is what is causing and why. What other tests can I run to diagnose source of problem. Someone mentioned Micro-Scope but not to sure.

Gigabyte 7VRXP v2
Socket A (462) w/Athlon XP 2000+
Southbridge : VIA 8233A(CE)
1GB (512 x 2 Mushkin DDR333) memory
nvidia G-force TI 4200
5 fan case
No OC'ing

Running Autodesk ADT/3D Viz 2005. No gaming

Thank you for any assistance!

 

[imported_Phil]

What power supply are you using? Is it a generic, or "no-name" brand? What wattage?

 

[tja111]

Antec atx12v (460W max) low noise model not sure which one. Not the illuminated model.

 

[imported_Phil]

If the RAM really is hot to the touch, then it's possible it's being run out-of-spec. What VDIMM (memory voltage in the BIOS) is it set to? Are you using SPD timings, or are they set manually?

Secondly, is the RAM running at 333Mhz, or 166Mhz DDR? If you're running it at 400Mhz (200Mhz DDR), for example, then you'd be overclocking it, which may be the cause of your problems.

Lastly, do you have another power supply you could test the machine with? A 300-350W should be fine for a quick test, and it's not uncommon for failing power supplies to do weird things to your machine

 

[tja111]

"What VDIMM (memory voltage in the BIOS) is it set to? Are you using SPD timings, or are they set manually?"

-Don't know, will have to look @ the BIOS. I know that I set everything to default when I built the system. Added the second memory module later.

"Secondly, is the RAM running at 333Mhz, or 166Mhz DDR? If you're running it at 400Mhz (200Mhz DDR), for example, then you'd be overclocking it, which may be the cause of your problems."

-How do I determine this? I am assuming I will have visability to this within the BIOS?

"Lastly, do you have another power supply you could test the machine with?"

-No unfortunately not. I bought the best PWR supply I could at the time. I have loaded the utility manager SW that came with the board to see if I am somehow under powered etc.

 

[imported_Phil]

Originally posted by: tja111
"What VDIMM (memory voltage in the BIOS) is it set to? Are you using SPD timings, or are they set manually?"

-Don't know, will have to look @ the BIOS. I know that I set everything to default when I built the system. Added the second memory module later.

It may be worth having a look at what the VDIMM is currently set to; wise advice from mechbgon is to raise the VDIMM to 2.7v, but if your RAM is already running hot then I'd advise caution on that at this stage.

[Edit] You say that one of the modules is too hot to the touch, and not the other one? That sounds like that particular module is being stressed beyond it's limits- it may be a PC2100 (133Mhz/266Mhz) module, not a 333Mhz (166Mhz DDR). Something to check! SiSoft Sandra (free download) can give you detailed information on your memory modules, including what speeds they can run, at what timings.

"Secondly, is the RAM running at 333Mhz, or 166Mhz DDR? If you're running it at 400Mhz (200Mhz DDR), for example, then you'd be overclocking it, which may be the cause of your problems."

-How do I determine this? I am assuming I will have visability to this within the BIOS?

Yes, you will see something usually labelled "DIMM Frequency" or similar. Sometimes "RAM Frequency" or "RAM Clockspeed". This should be set to the same as the FSB for your processor, which will be either 266Mhz or 333Mhz (133Mhz DDR and 166Mhz DDR respectively). I think you'll be looking at 333Mhz, but a look at the FSB Frequency will tell you that.

"Lastly, do you have another power supply you could test the machine with?"

-No unfortunately not. I bought the best PWR supply I could at the time. I have loaded the utility manager SW that came with the board to see if I am somehow under powered etc.

That's a pity. Does the monitoring software give you a readout for the voltage rails? You're aiming for a 10% swing either way from the stated voltage, anything more than that and you're looking at problems.

Lastly, what does the CPU temperature report (usually Hardware Monitoring)? Anything more than 55-60C in the BIOS and that's likely to be your problem right there. As a rule of thumb, 30-40C is "comfortable", 40-50C is "getting a bit too warm" and anything more is usually indicative of a heatsink mounting problem, insufficient thermal paste (or too much), or poor ventilation.

 

[tja111]

I think you found the problem.

Loaded the board utilities program that came with my board. What garbage by the way...Not to happy with Gigabyte

Anyhow, noticed the CPU temp was @ 61c

Decided to remove the fan and heatsink, found alot of dust and almost no thermal paste between heatsink and chip.

Bought a new fan/heatsink from some company called MadDog the (SilentChill K7) who comes up with these rediculous names???

Anyhow a little Arctic Silver, a new fan and heatsink and we are operating at a comfortable (though not optimal) temp of 46c

Also looked the memory. Using Mushkin PC-2700 512mb x 2 sticks.

BIOS has a setting SDRAM to SPD = enabled, then modified the SDRAM voltage to 2.7v

Now that we have done all this, and I realize I really don't like my Gigabyte board very much I think I am going to get a new board and chipset. Any suggestions???

Thanks!

 

[sarahl]

hi i ran memtest 86 through 8rounds of tests with zero errors. however when i switched it to bios-all i received 105 errors. all this occurred while running at stock speeds. is there cause for concern?