Hello everyone,
beginning of August 2019 i built a new PC with the AMD Threadripper 2950X and started to get freezes (and also crashes) more and more frequently and now i'm out of ideas what the problem could be.
My guess is that either the CPU might be faulty or something with the voltages is fundamentally wrong, since it got worse over time and i can reproduce a freeze or crash (sudden restart of the system) by running a computational heavy program (like wPrime) on all cores or have a lot of browser tabs open.
It is worth noting that running my PC the first two weeks was totally fine. Now it freezes every time i run wPrime. Since today, i also had a lot of trouble getting the PC to boot at all after it crashed while watching a stream. I even got a "Windows Automatic Repair" screen with crashes.
My build:
- Motherboard: ASRock X399 Taichi AMD X399 So.TR4 Quad Channel DDR4 ATX Retail
- CPU: AMD Ryzen Threadripper 2950X 16x 3.50GHz So.TR4 WOF
- RAM: HyperX HX430C15PB3K4/64 Predator, DDR4, 64GB (Kit 4x16GB), 3000MHz, CL15, DIMM XMP
- Graphics Card: 11GB MSI GeForce RTX 2080 Ti VENTUS 11G Aktiv PCIe 3.0 x16 (Retail)
- PSU: 1000 Watt Corsair HXi Series HX1000i Modular 80+ Platinum
- Storage: (2x) 500GB Samsung 860 Evo 2.5" (6.4cm) SATA 6Gb/s 3D-NAND TLC (MZ-76E500B/EU), (2x) 500GB Samsung 860 Evo 2.5" (6.4cm) SATA 6Gb/s 3D-NAND TLC (MZ-76E500B/EU)
More details via Speccy: speccy - Pastebin.com
Screenshots from AMD Ryzen Master (sorry for the language, but looks like i can't change it):
Other useful information:
- My BIOS of the motherboard are up to date.
- All of my BIOS options are default (i haven't even started any overclocking yet).
Things that i already tried:
- I tried swapping out the RAM with different modules and also tried out only having only one single module installed and also tried out the XMP profiles of the above listed RAM. My current modules are also listed in the QVL of ASRock (ASRock > X399 Taichi):
(suspicious is only that my RAM is listed as DDR4-2400 in the BIOS menu)
Are my voltages or frequencies of the CPU and RAM enough (see speccy-pastebin above) or do you have any other idea what the problem could be?
Thanks in advance,
Eigen
Thanks for the quick response John,
i basically tried out the two stick and four stick combination as described in the MB manual, as well as a single stick combination in the slot A2:
I will try out some combinations you mentioned and see what happens. Unfortunately due to work i'm only at home on weekends to test things out, so please be patient if my report on this might take a week.
Kind regards,
Eigen
The RAM Memory you have is listed under your Motherboard's QVL List. For that particular RAM, the Motherboard supports populating 2 and 4 DIMM sockets. So you RAM Memory should be completely compatible with the Rzyen and motherboard if you populated it under either 2 DIMM sockets or 4 DIMM sockets.
To eliminate defective physical RAM MEMORY run MEMTEST86 for about 2-3 times. IF it shows ZERO (0) errors than your RAM is probably good. You might want to run it before you go to bed because of the amount of RAM you have installed it might take longer than a couple of hours to run those many tests.
Do you have the latest AMD Chip set installed : ASRock > X399 Taichi
Any error messages when it freezes or crashes?
Try stress testing your CPU and GPU and PSU using OCCT and see what happens. If it crashes or passes. That is a good method in determining if you have a overheating or power or possibly defective hardware.
Just for troubleshooting purposes, DISABLE half the cores in your Threadripper. Run 8 cores instead of 16 cores and see if it crashes. You can use Ryzen Master, I believe to disable half the processor's cores.
Also run SFC /scannow in a elevated Command Prompt or Powershell. This is to check to make sure your WIndows OS is not corrupted or missing any core Windows files.
Finally you can always open a AMD WARRANTY REQUEST (https://www.amd.com/en/support/kb/warranty-information/rma-form ) and let them know of your symptoms and see if they believe the Processor needs to be RMAed or not.
Also open a ASRock Support ticket and asked them if it is possible your motherboard may be defective.
NOTE: In Windows 10, when you restart the computer 3 times before reaching Windows Desktop, it automatically starts the Windows Repair window.
Personally it sounds like a hardware is going bad. Especially if you are having a difficult time rebooting back into Windows Desktop.
When you need to reboot more than once to reach Windows desktop, are there any Motherboard Trouble LEDs , Codes, or strange BEEPS?
Thanks for the reply elstaci,
to respond to some parts of your answer:
- I can try out MEMTEST86 next weenend (i'm only at home on weekends) on my PC, but i highly doubt the RAM itself is faulty, since this is already my second set of RAM sticks i tested with my build.
- I don't think i had the latest AMD Chip Set installed. To be sure, i just installed the update from your ASRock link.
- I tried out disabling half of the CPU cores a month ago without any success. But since this was on my old set of memory sticks (which was not in the QVL of my motherboard), i will repeat the test.
- I just ran "sfc /scannow" on my PC and some errors popped up:
But after reading the log, those errors were all due to Windoes Defender PowerShell files, which are marked as false-positives according to here: https://support.microsoft.com/en-ie/help/4513240/sfc-incorrectly-flags-windows-defender-ps-files-as-... - so the OS should be fine.
- Regarding the errors codes on freeze/crashes. The ASRock motherboard is using "Dr. Debug", which is a return code that is shown on the motherboard: ASRock > FAQ . While rebooting after a crash or freeze i get a lot of "16" return codes, which is memory related. But i also got a "00" code (CPU related) and SATA-related return codes too, hence also my assumption, i could be hardware related. As far as i remember no beeps were heard during the whole booting process. But all components like fans are running like normal in case of a freeze.
Further open tasks that i have to test next weekend is the power supply stress testing (OCCT). I'm very curious about the result since i have a lot of hardware hooked up.
Kind regards,
Eigen
