Processors

[obeliskag]

Hello,

Having an issue with my 5950X system. When the CPU has a sustained heavy load, the system is crashing. The CPU is not overclocked and temperature under load is ~56°C. (280mm AIO). DOCP enabled in BIOS.

Checking Windows Event Viewer, Event 18 error messages of concern are:

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 27

And similar error message:

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 1

System components are:

• CPU: AMD Ryzen 9 5950X
• Motherboard: Asus Prime X570 Pro with latest BIOS 4201
• RAM: G.Skill Trident Z Neo, 16GB x 4, 3600Mhz, 16-16-16-36 (On the QVL list for the previously install 3800X)
• Video: Gigabyte Aorus GeForce RTX 3080 Ti XTREME 12G
• Store: Samsung 970 Evo Plus 1TB
• NIC: Asus XG-C100C
• PSU: Corsair RM850

Do I need to RMA my processor, or what is required to resolve this issue?

Thanks for your help...

[obeliskag · Accepted solution]

Did some more troubleshooting and made the following changes:

• In BIOS, changed "Load Line Calibration" from Level 3 to Auto (In the past, I had experimented with ClockTuner for Ryzen to reduce temps and power.)
• In Windows, I had disabled Specter protections - I turned these back on

I am monitoring the situation, but so far the system appears more stable.

[s4a]

Same error. Also ASUS X570 motherboard, also GSkil memory. Crashes with  CPU not running anything but with more memory in use the more often. With 128 GB crashes every second day

Tried ASUS support asking may be BIOS update is needed they still take a month to respond. 

May be AMD knows the answer?

[obeliskag · Accepted solution]

Did some more troubleshooting and made the following changes:

• In BIOS, changed "Load Line Calibration" from Level 3 to Auto (In the past, I had experimented with ClockTuner for Ryzen to reduce temps and power.)
• In Windows, I had disabled Specter protections - I turned these back on

I am monitoring the situation, but so far the system appears more stable.

[ryzen_type_r]

You have in fact overclocked the CPU's Infinity Fabric, as fabric speed is locked to memory speed (unless decoupled in BIOS), and 3200 is the max officially supported memory speed.  So there are no guarantees for anything above 3200 - even if the memory itself can do 3600, the fabric may not be stable so this will require more trial and error on your part.  Sometimes you win the silicon lottery, sometimes you lose.

Additional things you can try to improve stability:

- Run memory speeds at 3200 or lower, just to see if fabric speeds are the issue
- Bump up Vsoc and/or Vdimm voltages slightly
- Run 2 sticks only (some DIMMs just don't do well in quads)

[Matt_AMD]

Has your memory been validated to run at 3600Mhz, using 4 DIMMS? Please can you provide the memory model number.
I can see that your system memory is overclocked.
Running 4x16GB sticks at 3600Mhz might be causing your issue.
Put the memory back to 2133Mhz and see if the same issue occurs, or try removing two DIMMS to see if you get stability at 3600Mhz.

[obeliskag · Accepted solution]

Regarding the memory kit, the system previously had a Ryzen 7 3800X and the G.SKILL Trident Z Neo F4-3600C16Q-64GTZN is on the "Memory QVL for 3rd Gen AMD Ryzen Processors X570". However, this memory kit does not appear on the "Memory QVL for AMD Ryzen 5000 Series Processors_X570" list.

I purchased the processor on launch day, but the 5000-series QVL memory list for my motherboard was not published for some time after that. All indications prior to launch, was that the 5000 series IO die was equivalent to the 3000 series, so I would not anticipate an issue with RAM that is on the 3000 series QVL.

The issue is intermittent and only exhibited with a specific application. After running more memory tests and benchmarks, I am suspicious this is a specific application issue. However, it is surprising to see an application crashing a system to hardware reset in 2021. I am also wondering if my PSU is not performing to specifications.

[buildorbust]

this beacause bad cpu quality.

You have 2 options:

1. RMA.

OR, if rma not possible

2. Increase TDC limit from 95A to 115-125A.

-----

many people encounter random reboot on stock settings from november 2020 till now.

YOU ARE NOT ALONE!!!

[s4a]

Changed memory settings to 2300 from 3600. Crashes

Changed power supply. Crashes

Changed all BIOS settings to recommended in all forums. Crashing

Asked ASUS if they experienced this error. They sent me RMA for repair motherboard without answering my questions.

Will try to buy another motherboard and try before sending them to see because doubt they will "repair" anything as crashes happen sometimes twice a day, sometimes ones a week

Suspect this is AMD problem. Always had problem with it. My Intel rig does not crash already for almost a decade running day and night