Resolved Mac Pro 2013 hard Crashes / CPU Panic

[-paradox-]

Hi!

My MacPro 2013 is giving me hard crashes every half an hour or so (CPU Panic most of the times).
I'm running 2 separate partitions of Mojave and I'm getting these crashes on both of them (so not software related...).
When it happens Keyboard and Mouse stop working and if any sound is playing at this moment then a junk of it is getting repeated (but that might always be the case?!). After 5secs or so the mac shuts down.

According to the apple-hardware-test ("d" on startup) everything is ok, obviously it isn't

I've installed 64GB Ram when I bought it in 2014 and a Samsung SSD 970 EVO 2TB in December 2018, but everything was working fine until like a month ago or so (I've attached a passiv cooler to the SSD, when I installed it). The crashes came out of nowhere...

I do know a lot about Mac / PC basis but when it comes to such kind of problems I'm completely lost... That's why I would appreciate it, if someone could take a look at the 3 crash reports I've attached and let me know if there's something in there that might point to a specific problem.

If not, what would you start with:
-unplug all unnecessary devices?
-replace the SSD with original Apple SSD?
-replace the RAM with original Apple RAM?

Thanks in advance!

 

[keysofanxiety]

Hopefully somebody else will have a better idea what's wrong but I'll chime in my $0.02.

There's no consistent process to the crash reports (Windowserver, sandboxd, loginwindow) which points to general system instability. The 2013s have an high failure rate and a repair program for the D500 & D700 GPUs - you have the D500, so that's always a potential cause. However in my experience of graphics failure, the crash report tends to be a little less ambiguous.

Based on that, we'll start with the RAM. Firstly to reassure you: you've got the correct speed RAM installed and it's ECC. Can you try install MemTest x86 to an external USB and boot into that on startup to test your RAM: https://www.memtest86.com/technical.htm

I'd recommend the above rather than swapping the RAM as leaving this overnight will show if there are any problems with the RAM. It's an industry standard application and is extremely unlikely to miss any issues.

If that comes back clean my only other guess is the graphics based on historical hardware failure, though again I want to start with the RAM because graphics failure is usually a little more obvious in the crash logs (in my experience).

 

[-paradox-]

Hi!

Thanks a lot! I‘m running memtest right now. I‘m not sure how to interpret what it‘s already showing me (in red):
A couple of ECC errors all on channel 1 in basically every Test it has run (time only 50mins).
But I have 0 errors on the right hand side of the screen (meaning they got corrected because of ECC, is that correct?!).
I‘ve read somewhere else that ECC errors might also point to a faulty processor?!

 

[-paradox-]

Hey keysofanxiety!

Even though it seems you're not following this thread anymore, i just wanted to thank you again!
All I got in MemTest86 were ECC Errors in slot 1 (that got fixed, so no "real" errors).
I took out all the RAM-Sticks, cleaned them and stuck them back in (but switched the one that was giving me ECC errors from Slot 1 to Slot 2). Now this stick doesn't even get recognised by the system anymore, but I didn't have any crashes since then. Will get a replacement for the faulty RAM and then I'm back on 64GB and happy about a stable system
Thanks a lot!