Help on hardware diagnostic

[Gfurst]

Hey people i'm having the damnest hardware crash ever, not because its really bad, but because its really random.
Here is another thread with more information on what i been through http://ubuntuforums.org/showthread.php?p=11065750

Anyway, my problem usually happens on heavy load but not always, the pc crashes, i got some weir distorted pixels around the screen( gonna get a shot as soon as i can), sometimes others thing kinda run halfway for a second or so( like music) and then i have hard reboot.

So i'm trying every possible test to rule out the good guys and not have to around buying unnecessary parts. I usually use linux, but also have windows on another driver, so i got your BurnIn software to help me diagnose my problem.

As it happens, it initially showed no errors, but with cpu+ram+2dvideo+3dvideo test it crashed after a few minutes.
So i though "finally, something that will help me", after the hard restart i run every test involved individually, but no luck, none showed any errors.
Also i couldn't get the crash to replicate again( argh!).
After a 2dvideo+ram test, i got about 23 errors on my video memory, but i read in this same forum that video memory usually causes graphic corruption, but not system crash( not in these same words).
Even so i though "yeah, finally the video is the culprit", but still i can't these errors to replicate themselves, even with other specific video memory testing i can't find anything.
I'll be damned, i guess my pc doesn't like me anymore...

So anyway, i hope you can help me find what could be causing this, the help is very appreciated. Also i have all the log files available( except the one which crashed), ask me which one is better to post and i will happily do it.

[David (PassMark)]

We also have a version of BurnInTest for Linux, so you could have done some testing in Linux if you preferred.

If it crashes in Linux and Windows in a random fashion, then problem is most likely hardware.

It is time to start swapping out some hardware. Start with the video card. Either borrow a friends video card, or buy a cheap one for testing off Ebay for a few $$ and try it out.

If you have 2 sticks of RAM, then remove one at a time, and see if stability improves.

If RAM and video card aren't the problem, then it is time to start thinking about a new motherboard. You said the machine is 4 years old. So failure at this age isn't unusual. PC's aren't built to last 10 years.

[Gfurst]

Oh, i didn't saw this linux, nice touch , I'll test it right now.

It crashes in both OS, i don't have any spare parts, I live in Brazil so E-bay is kinda non option for me. But i just saw a reasonable card is not so costly.

Like i said, i was wishing to detect the defective part before starting to buy spares.
I was suspecting the motherboard too, getting a lot of weird bugs. Also I've just replaced my PSU which fried out not long ago, i was afraid it could've damaged other parts.

Truth is, i'm planing building a whole new PC from scratch, but thats only gonna be possible within a few months, so i want to salvage this one as long as i can.
High end parts are really expensive here in Brazil.

edit:
Oh and one more thing, i ran four complete test on full load trying to trigger the crash and it didn't happen, thats a whole hour on full load, how could a faulty hardware supported that?

[David (PassMark)]

Swapping out suspect hardware is a big part of PC repair. In many cases it is not possible to know the exact root cause of a problem without kernel debuggers, bus traces, and a team of engineers. For example, if the problem was EMI, bad capacitors, or dry solder joint. You'll never find this out by running software.

If you can't swap your video card yourself, then maybe just take it to the local PC repair store.

[Gfurst]

Wow, i just ran a brief test in linux, nice job!

Within linux i can keep i better eyed on the monitors, the CPU temp fastly rose from the cool 31°C it usually keeps to a constant 46°C, the GPU went from 52°C to 56°C, everything normal so far. The RAM however had already 66% used by cache leaving not much to be tested from.
About the 2D test, it doesn't work as in windows with the video memory test purpose, right?
I've guess you could add Ubuntu to the successful distro list, here is a detailed log:

BurnInTest v3.0 (1001) Linux
Logging detail level: Result summary

*** Unlicensed Trial Version ***
******************
SYSTEMN SUMMARY
******************
Linux 2.6.35-30-generic
1 x Intel(R) Pentium(R) Dual  CPU  E2140  @ 1.60GHz
2.0 GB RAM

GENERAL
System Name: gfurst-desk
Motherboard Manufacturer: ASUSTeK Computer INC.
Motherboard Model: P5LD2-X/1333
Motherboard Version: Rev x.xx
Bios Manufacturer: American Megatrends Inc.
Bios Version: 0204   
Bios Release Date: 12/25/2007

CPU
CPU Manufacturer: GenuineIntel
CPU Type: Intel(R) Pentium(R) Dual  CPU  E2140  @ 1.60GHz
Speed: 1200.0 MHz
CPUID: Family 6, Model 15, Stepping 13
Socket:  LGA 775
Physical CPU's: 1
Cores per CPU: 2
Hyperthreading: No
CPU Features: MMX SSE SSE2 SSE3 PAE 
Cache per CPU package:
L1 Cache: 64 kB
L2 Cache: 1024 kB

MEMORY
Total Memory: 2012 MB
Available Memory: 129 MB
Memory Devices:
DIMM0 : DDR2, 2048 MB, 333 MHz (3.0 ns)
DIMM1 : Unknown, No Module Installed, Unknown

**************
RESULT SUMMARY
**************
Test Start time: Wed Jul 20 02:28:44 2011 
Test Stop time: Wed Jul 20 02:38:47 2011
Test Duration: 000h 10m 03s 

                          Test Name   Cycles   Operations      Result  Errors   Last Error
                        CPU - Maths   388      49.729 Billion  PASS    0        No errors
                       Memory (RAM)   188      30.254 Billion  PASS    0        No errors
                        2D Graphics   341      11.292 Million  PASS    0        No errors
                        3D Graphics   3602444  1.346 Billion   PASS    0        No errors
TEST RUN PASSED

*********************
SERIOUS ERROR SUMMARY
*********************

Anyway, back to my problem. I know eventually i'll have to test with other parts, but it seems useless yet as i can't even trigger the crash when i want to.
Like i've said before, running a heavy load test on windows, i got the crash under 3 min, I hard restarted the pc, run the test singularly, but everything passed without errors. So i went for the crash again, heavy load test, and the damn thing ran 4 times without a glitch.
Sure this could be anything, even the motherboard, but i would think errors within that would be more apparent. Also maybe the PSU i got is not that good, failing to supply the necessary power in key heavy load moments.

I'm sure you've more experience in this than me, i was hoping you could help me a bit more. Thanks!

[David (PassMark)]

There is no magic wand.