Win7x86 Crash Dubbed "Disco Screen"

[TofuCrew]

Here is a video of the dreaded "Disco Screen" : Win7x86 Crash. Dubbed "Disco Screen" - YouTube

I've been dealing with this issue for 3 weeks. Now it's time to solicit the help of you fine folk

History: We order about 25 ACER VM6630G computers a month. As of December 2013, the manufacturer made a change- they "upgraded" the CPU from an i3-3220 to an i3-3240. Since this change we've seen a failure we've dubbed as the "Disco Screen" on about 5 of 28 computers (~18%). I'm not positive this change is the smoking gun as we may have a (unconfirmed) case of an older, i3-3220 machine, exhibiting the issue.

Symptom: The machine will become unresponsive, displaying a rainbow of rapidly changing solid colors, followed by shutting off. In this time, there is no BSOD. There is no kernel dump. Just dance party.

I've had the following failed machine, sent from a customer, on a bench since 1/28/2014
Machine running:

1. Acer VM6620G
2. - i3-3240, 4GB RAM, 500GB hard drive
3. Win7x86 w/ Windows updates until 11/12/13
4. National Instruments LABVIEW
5. Communicating with a Galil Motion Controller over Ethernet Via Intel Gigabit CT desktop adapter
6. Aquiring Point Grey Firefly Firewire camera feed via NI Vision Acquisition module through Point Gray Research Firewire PCI card (ACC-01-1000)

The following are my steps taken so far:

Tuesday, 01-28-2014: 11:50am

• Started PC on bench running as close to failure conditions as possible.

Friday, 01-31-2014 8:00am

• Came in to find machine off.
• Unexpected shutdown on 01/30/2014 @ 4:51pm (via event viewer)
• - Ran for 2 days, 5 hours
• Set up to run again over weekend

Monday, 02-03-2014 7:30am

• Came in to find machine off
• Unexpected shutdown on 02-02-2014 @ 12:20am
• - Ran for 1 day, 16 hours
  8:00am
• Started burn in software (Furmark/prime95) to test CPU/GPU

Tursday, 02-06-2014 10:00am

• Stopped burn in -Total burn-in run time: 3 days, 2 hours
• returned machine to previous test conditions (did not restart)

Monday, 02-10-2014 7:30am

• Came in to find machine off
• Unexpected shutdown on 02/09/2014 @ 10:34pm
• - ran for 6 days, 3 hours

Tuesday, 2-11-2014 8:30am

• Installed Intel Video Drivers (Win32_15338)
• - Old driver rev- 9.18.10.3071 Date: 03/19/2013
• - New driver rev-10.18.10.3345 Date: 10/31/2013
  
  3:00pm
• Started Intel Extreme Tuning Utility CPU Burn
• Furmark 1.12 burn-in test w/ settings:
  1. Dynamic Background
  2. Burn-in
  3. Xtreme Burn-in
  4. Post-FX
• Logging Temperature to file

Wednesday, 2-12-2014 7:30am

• Started Prime 95 w/torture test (Previous burn-in still running. Burn, baby, burn.)
  
  3:15pm
• clicked on System clock in taskbar. Video driver seemed to crash, leaving the display @600x400, focused on calender.
• system ran for 24 hours
  
  3:45pm
• Started verifier w/
  • Create Custom Settings (For code developers)
  • select individual settings from a full list
  • enabled everything but low resouce simulation
  • selected all drivers but Microsoft
• Starting NI Software immediately crashes system with nipalk.sys Stop: 0x000000c4
• Tested this 3x
  
• unchecked NI drivers from list
• software now runs. left for the night

Thursday, 02-13-2014 (today)

• Machine still running when I came in (7:30am)
• Went to use SF_Diagnostic_tool, came back, machine is off.
• unexpected shutdown @ 8:16am

At this point, it's taking too much time to diagnose and/or fix. Steps I'm going to take I'm going take from now: going to take an image of the machine then put on a factory default ACER image and see of that will run without failure. If not, I'll take out all cards, then try to work backwards. I'm starting to think it might be a windows update that I installed in November of last year? Honestly, I don't know....

 

[Arc]

Driver verifier caught National Instruments LabView as failing.

BugCheck C4, {f6, 91f4, aff55270, 863b13c0}

*** WARNING: Unable to verify timestamp for nipalk.sys
*** ERROR: Module load completed but symbols could not be loaded for nipalk.sys
*** WARNING: Unable to verify timestamp for niimaqdxk.dll
*** ERROR: Module load completed but symbols could not be loaded for niimaqdxk.dll
[COLOR=Red]Probably caused by : nipalk.sys[/COLOR] ( nipalk+583c0 )

Followup: MachineOwner
---------

And it is old.

8634c000 863e8000   nipalk   T (no symbols)           
    Loaded symbol image file: nipalk.sys
    Image path: \SystemRoot\System32\drivers\nipalk.sys
[COLOR=Red]    Image name: nipalk.sys
    Timestamp:        Thu Jun 30 00:10:51 2011[/COLOR] (4E0B71B3)
    CheckSum:         00090DD0
    ImageSize:        0009C000
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4

If possible, update it. NI Measurement Studio - National Instruments

Also, dont Let it auto start with windows. Remove these items from startup:

NI Error Reporting    c:\progra~1\nation~1\shared\nierro~1\nierse~1.exe    Public    Common Startup
NI Update Service    "c:\program files\national instruments\shared\update service\niupdateservice.exe" -startuptask    Public    HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run

But it may/might not be the true cause behind the disco lights. Do some more hardware checks.

Test your RAM modules for possible errors.
How to Test and Diagnose RAM Issues with Memtest86+
Run memtest for at least 8 passes, preferably overnight.

If it start showing errors/red lines, stop testing. A single error is enough to determine that something is going bad there.

Check if the Power Supply Unit (PSU) supplying adequate power to the computer or not.
eXtreme Power Supply Calculator
Also let us know the wattage of the PSU.

We cannot test the GPU here, because it is integrated; a component of motherboard.

Let us know the results.