CPU: Ryzen 5 1600

MB: ASRock AB350M Pro4

Memory: Kingston HyperX 2x8Gb DDR4-2400 (HX424C15FB2K2/16)

Graphics: MSI GeForce GTX 1050 Ti GAMING X 4G

PSU: EVGA SuperNova G3 550W

I've tried running Memtest86 in Single CPU mode with each DIMMS in all slots and each time the machine either freezes or reboots (no actual errors reported). Strangely when the machine has been off for a while, the freezes/reboots take longer to manifest themselves but once the machine has had a freeze, other freezes happen usually in the 6-8 minute period of Memtesting.

In Windows 10, starting from a "cold" PC, I can sometimes use it for 2 to 3 hours but from then on it will freeze randomly - sometimes after minutes, sometimes after an hour or so.

Have checked the CPU & motherboard temps after a freeze and they are usually sitting at a max of around 50 celsius for the CPU and about 40 for the MB.

I'm wondering if anyone has any ideas about whether this issue sounds more likely to be memory, motherboard, or CPU related so that I know which one I should look to RMA first. Unfortunately I haven't found anyone else with a DDR4 capable machine to test the memory separately.

Any thoughts/advice greatly received!

If your case is cramped and airflow to the CPU socket area and surrounding components is bad that could easily cause freezing and restarts as the power circuitry triggers thermal protections. Usually a fan aimed down onto that area sorts things out. 

Just reading back what I wrote originally I should have probably pointed out that 95% of the time freezing is a problem rather than rebooting and when the machine freezes sometimes it does carry on processing some stuff for a short time (e.g. once it froze when playing a video but the sound carried on playing for about 20-25 seconds before that froze too). Is that still likely to be a problem with VRM overheating as reboots seem to be the biggest symptom from things I've read rather than freezing? (It does sound like a good potential for the cause of the issue so not dismissing it but figured I'd ask the question :) )

Also just a few quick questions related to the overheat theory:

* I'm assuming that the Overheat protection in the BIOS only refers to cutting out when the CPU overheats and not the VRM components as I did a brief test with that disabled?

* Would it be normal to not see high temperatures in HWMonitor/BIOS HW Monitor screen following a freeze? (i.e. would the temperature spike and drop so quickly that you wouldn't see any evidence there?)

* Is there a particular temperature which the thermal cut-out triggered by or is the cut-out monitored by other factors? 

Thanks again everyone for their replies so far - it gives me great relief in having people to bounce ideas off and makes me more hopeful of solving my issue before I lose all my hair :)

Experts shall understand it, it is better to give to service center.

So, here goes.

Overheating can cause any number of issues from slow, choppy performance to hanging (brief freezing that resolves itself) to full thermal shutdown and being unable to power on the PC again until the components have cooled down.

* I'm assuming that the Overheat protection in the BIOS only refers to cutting out when the CPU overheats and not the VRM components as I did a brief test with that disabled?

Overheating protection (the setting) applies only to the CPU. VRM temps are not monitored by the UEFI. 

Yes, it would be normal to see all temps reporting as OK in monitoring software. As I said above, these temps are not reported. I would imagine because it would require a thermal sensor for each of the VRMs on the board which in the case of a 12 phase power design could be very expensive. Thermal shutdown/issues here are detected not by monitoring temps but by voltages going out of tolerance due to heat. Once the power controllers (chips on the board) detect this they trigger safeties to protect the system.

* Is there a particular temperature which the thermal cut-out triggered by or is the cut-out monitored by other factors?

We've placed a nice cougar 120mm fan right on the vrms, and the cpu cooler isn't stock also, everything is default clocks, and the vrm heatsinks aren't hot to touch. 

Installing the latest bios, defaults, a2, b2 for ram, a1 b2 for ram, now on a2 with one dimm, installed latest amd chipset drivers, fiddled around with everything in windows, gave more voltage for the ram dimm, still that same crash since day one of purchasing it, turning off overtemp protection in UEFI also didn't change anything. Ryzen power plan also didn't change anything. 

Reseated cpu power, 24pin power, ram, reset bios numerous times.

Things to try still:

Beta Bios 4.63

Firmware update for Crucial m500 (doubt this is affecting it, as it ran flawlessly in my ps3)

Reseating of CPU (maybe?)

But nontheless, from my experiments I doubt it is temperature related...Wish you luck and please update us here! so many asrock ab350m pro4 owners being plagued by similar crashes

https://www.reddit.com/r/Amd/comments/78pc3x/new_pc_amd_ryzen_5_1600_32_asrock_ab350m_pro4/" rel="nofollow - https://www.reddit.com/r/Amd/comments/78pc3x/new_pc_amd_ryzen_5_1600_32_asrock_ab350m_pro4/

I've tried - I submitted an ASRock Technical Support form a few days ago but I've yet to hear anything. I'm guessing that perhaps I've fallen into the same trap as Kingston Technical Support who are all unavailable from the 14th to 21st due to Chinese New Year.

So figured in the meantime while I wait I'd ask people who might have been here before.

Thanks for taking the time for the explanations Xaltar - great info and definitely helped me understand!

Unfortunately, although I think the issue might be partly thermal related, I'm not sure that's the full story. I did a test last night with the USB fan blowing across the components and although it lasted a bit longer, it decided to freeze eventually. 

I'm starting to think, based on a lot of other reboots of freezing, that perhaps either its not VRM heat related or that the VRM is very picky about the temperature and is tripping at lower temps than it should.

I'll probably be getting hold of the supplier of the motherboard to discuss options - either try a replacement or perhaps choose a different brand/model as I've read stories about boards that start off ok but fail after a period of months.

I actually wonder if ASRock are doing a quiet "recall" on this board as, although it was quite common in the webshops in NZ when I purchased it, it now appears to have disappeared from almost every shop and the couple that do have it now have it on "special order". (Conspiracy theories are great huh? :D)