Computer crashes randomly with multicolored screen (usually a garbled, finely tiled mess of repeating pixels that were on the screen before the crash). Happens regardless of OS (that means Windows and Ubuntu).
It's an AMD motherboard with NVIDIA 6150SE graphics, 3GB DDR2-669 ram, AMD Athlon 64 X2 6000+ CPU. What's wrong with the computer? Here are the tests I've performed:
1. Swapped monitors.
2. Reformatted hard drive and installed fresh OS (multiple)
3. Ran memtest86+ 4.20 for nearly 3 passes (took hours)
4. Ran prime95 stress test in 64-bit, hi-ram SMP torture mode: about an hour
5. Changed BIOS settings to failsafe preset, disabled cool'n'quiet, SMART drive, acpi, all speed tweaks, and then locked fans to max, etc -- then did the reverse and redid all tests
6. Booted from SD card with acpi=off and video=vesafb vga=792 (to bypass all gfx acceleration and nvidia module loading; I can't remove the onboard GeForce chip)
All tests come out clean but spontaneous hard-freezes persisted, with the screen either glitching out with tons of random tiled colors (or pixel repeats) or freezing to black. No responses or interrupts were accepted at this point -- if the fans weren't set to max in the bios, they'd stop, too. Occurs regardless of activity or idle.
At first it seemed to me that it was a VRAM corruption problem in the GPU -- I've seen it before. Or perhaps a RAM problem, but all tests checked out. Temperatures stayed under 70C during use (including testing when they shot up from 40-ish). No overclocking was ever performed on this computer -- in fact, hearing terrible things about the nvidia chipset, I actually had applied a limiter to the GPU core freq (this limiter was accidentally removed around the time the crashes dramatically increased in frequency about 8 months ago -- but could be coincidence).
No new hardware installed. Here are my next best guesses, but they're kind of random and some insight into whether the other components could still be going bad is appreciated.
1. motherboard (aww, #@!^$ -- please no!)
2. PSU (which would be odd since it also happens when idle with acpi on and low fan)
3. Flow joints and/or thermal jelly around CPU/GPU
4. ...RAM or GPU, still, because my tests were inadequate.
Any suggestions or input? I really wanna know what to do with the poor rig, considering it was $700 new just a few years back and still races through its tasks... before it freezes.