0

I just got a new computer that far exceeds the system requirements for the game I was playing. The computer was not hot at all, but it crashed two different times with a WHEA_UNCORRECTABLE_ERROR (124) in a short amount of time playing. I saw a couple things in the .dmp file that seem to point to the processor, but I am not sure and would like some assistance interpreting the file and deciding what to do next. What are the relevant parts of the .dmp to share. What personal information could be in the file, and how do I avoid sharing that?

Thank you

oxydendrum
  • 1
  • The best test for hardware faults is [MemTest86](https://www.memtest86.com/). Leave it running for many hours, even overnight. – harrymc Apr 09 '21 at 20:13

2 Answers2

0

A WHEA_UNCORRECTABLE_ERROR (124) is a fatal hardware error.

Get the manufacturer's Hardware Test App and test the hardware (all aspects) .

If no support from the manufacturer, get memtest86.exe and the drive manufacturer's disk test app and test all memory and the drive.

If not the memory or drive then it is the motherboard including CPU that will be the most likely candidate.

A defective PSU might cause this error but I would look the errors above first.

John
  • 46,167
  • 4
  • 33
  • 54
0

You can either share the dump completely or not at all. It cannot be shared partly.

What’s in it depends on what type of dump you have: A minidump (could be several megabytes) or a full memory dump (usually gigabytes). A full memory dump is just that: All memory of everything that’s running on your PC, which could be full of private data like credentials. A minidump (technically “Small memory dump”) contains only the most essential information (bugcheck data, CPU registers, error structures if any) and will usually not contain private data.

That all being said, using WinDbg, you can convert a full dump to a minidump (quoting this answer on Super User):

A dump can be converted with WinDbg:

  1. Open the full dump
  2. .dump c:\debug\dumps\small.dmp

You can automate this task by using cdb instead of windbg and pass commands via the -c "<command>" command line switch, e.g.:

cdb -c ".dump c:\debug\dumps\small.dmp ; q" -z c:\debug\dumps\big.dmp

The q ensures that cdb quits after re-dumping.

Only instead I’d recommend using WinDbg preview from the Microsoft Store. It does not come with cdb.

With a 0x124 bugcheck (WHEA_UNCORRECTABLE_ERROR) it is most likely caused by failing hardware. Since you were pushing your PC to the limits, maybe the power supply was insufficient.

If you want quick help, you can share the results of running !analyze -v in WinDbg (it can take a while).

The bugcheck parameters indicate that a Machine Check Architecture error occurred. These errors are raised by the CPU when stuff goes wrong.

The parameters further hold the contents of the MCA_STATUS register that contains information about the error: 0xbc800800060c0859. According to the preliminary manual for AMD Zen 3 CPUs (the closest thing available), this means the following:

  • “The error was the result of attempting to consume poisoned data.”
  • The error was reported because reporting was on for “A parity error was detected in an LDQ entry by any access” (LDQ = “Load Queue”)
  • The error occurred while trying to fetch instructions from L1 cache

This means your CPU is malfunctioning. It could be damaged or it could be receiving insufficient power.

You can try stressing your system with Prime95 and/or Furmark to determine if it can handle full load.

You should use Memtest86 to determine whether all RAM is okay.

Since your system is new, it should be possible to replace components on warranty.

Daniel B
  • 60,360
  • 9
  • 122
  • 163
  • I doubt I was pushing it to its limits. I have an AMD 5600x, RTX3070, 16 gb 3600 ram and I was running Destiny 2, which has far lower recommended requirements. – oxydendrum Apr 09 '21 at 18:26
  • So you have a very new graphics card. It requires tremendous amounts of power. Are you _sure_ your PSU is up for it? // Lower requirements doesn’t mean it won’t push your PC if you turn up all the details. – Daniel B Apr 09 '21 at 18:37
  • It's a 650w PSU. – oxydendrum Apr 09 '21 at 18:43
  • FAILURE_BUCKET_ID: 0x124_0_AuthenticAMD_PROCESSOR__UNKNOWN_IMAGE_AuthenticAMD.sys showed up in both .dmps – oxydendrum Apr 09 '21 at 18:46
  • That is your CPU failing, so likely a motherboard needed. – John Apr 09 '21 at 18:52
  • Don't try to interpret yourself. A MCE can only be interpreted using the CPU architecture manual. Also, 650 W != 650 W unless it's a premium PSU. Also not if it's an old PSU. – Daniel B Apr 09 '21 at 18:55
  • How do I test if the power was insufficient? – oxydendrum Apr 09 '21 at 19:05
  • Heres the analysis from WinDbg: https://drive.google.com/file/d/1izG-FFt6ryOgbstPhrz64kqU-5c8DQo2/view?usp=sharing – oxydendrum Apr 09 '21 at 19:14
  • https://www.tomshardware.com/news/nvidias-rtx-3000-power-supply-requirements-PSU-shortage-2020 My PSU fits the requirements for 3070 listed here. – oxydendrum Apr 09 '21 at 19:16
  • Don't post Google Drive files that require authentication. – harrymc Apr 09 '21 at 20:10
  • I thought it was set to "anyone with the link." fixed. – oxydendrum Apr 09 '21 at 20:13
  • I updated my answer. In the future, please add textual information to your question inside a code block. That way, you won’t have trouble with permissions and your question will be self-contained and complete. – Daniel B Apr 10 '21 at 14:06