Super rapid unwanted shut down

[gn842a]

Just over four years ago I built an AMD/ASUS mobo desktop and have recently, as in over about four months, had a few incidents of super rapid crash. Basically the screen will freeze and then show a jagged static pattern (kinda funky like the old Outer Limits lead in) and then the whole computer shuts down, reboot itself, and everything is fine. The shutdown is so fast that I couldn't possibly photograph it with the smartphone. I'm pretty good about data backups so if this is the end of my desktop I'm prepared to build another one. But I would like to know what component is screwing up.

My hunch is that it is the mobo or CPU. My hypothesis is that if it were in the SSD I would get a BSOD and other signs of operating system failure. Istead, the system works pretty good: after the reboot the appropriate systems say there was an unexpected shutdown and they all do what they are supposed to do in terms of providing access to files that closed, etc.

The PSU is a thermaltake by no means lower end. (judging from these fora I'm supposed to move to Seasonic in the future). I've had plenty of PSU failures in the days I bought cheap desktops from Computer Renaissance, no such problems with this build, but this is a new kind of symptom I haven't seen before.

My initial hunch was that the system was overheating. I do some on-line gaming (am not ultra hard core) and have noticed CPU getting up to 73C. So I've been more mindful about setting the fans to max and when I game with them set up at high levels the CPU temp shows around 52C. But I can't say that these incidents happen *only* when gaming, and I only have two or three such incidents so far, so hard to generalize. The other thought is it might be a graphics card issue? I used this computer for a couple of years with no graphics card (one is built in to the CPU) but at some point got the R9-380.

AMD 10-5800k in ASUS F2A85-VPRO
16 gigs DDR3
AMD R9 380
Samsung SSD 850 Evo 500 gigs
Thermaltake 850W

I can't really recall whether the thermaltake is new or not. I run a second nearly identical build and may have swapped the units a year ago, when trouble shooting the upstairs (second) desktop. I suppose I should also mention that my home has a fairly antique 60 amp power box from the street but it has never been a problem (in particular, my stove and dryer are gas, so peak load is probably the refrigerator or when my girlfriend runs her hairdryer--the bathroom and the fridge are both on different circuits). I've had some flickering lights in heavy windstorms. Still, when we have out and out town wide power failures, which happens two or three times a year, the computer shut down is immediate with no freeze or weird screen display. So I'm not sure I want to try to trace this to the circuit box.

Any thoughts appreciated.

Greg N

 

[Rookie_MIB]

Well, 73C is very close to the tjMax for the A10-5700 series CPUs which is 74C. At that point the computer could be shutting down in self-defense. I would consider either a larger CPU cooler, or giving the cooler a good cleaning and blow-out along with the fan.

To be honest though, you could very easily get a nice Ryzen build for $350 or so (R5-1600 hex core, B350 motherboard, and some gigs of DDR4) and it would be hands down WAY better than what you have now. And you could keep the R9-380. That would be a solid gaming rig.

 

[gn842a]

Thank you for that fast answer. Your information about the TJMax of the chip is a good clue. There is, furthermore, a problem in that speedfan temps on CPU show 15C cooler than Speccy. Cortemp shows temperatures ranging from 0C to 7C and that's simply not possible. I could put my vegetables in there.

So there is a big question mark on the accuracy of these programs. I have an infrared gun I could use to shoot the CPU but it's covered up by the Noctua heat sink. The Noctua is about as big as I can fit in there.

I do blow out the computer twice a year, and have it twelve inches off the floor, and it's pretty much immaculate. I have seen what dirty PC interiors look like and we're not anywhere near that.

I will look hard at your build suggestion (have been interested by the Ryzen, don't know why I use AMDs really, but they seem to do the trick). But I have to point out that the costs are much greater than one might suppose, because if you change the mobo you need a new OS, and along with that goes various software installations that have to be re-done and in many cases re-paid for. And one can't be sure that the fancy heat sink will port to the new mobo. So it is in my interest to keep this thing going.

 

[Rookie_MIB]

Not necessarily (new OS). I had a Win7 64bit Pro installation on my main rig, which was running my Asus P7F7E-WS/Xeon X3470 build. I downloaded the AMD motherboard drivers to a USB stick, uninstalled the nVidia drivers, the Asus drivers, shut down the computer, put in the new motherboard/CPU/RAM in, started it back up. It found almost everything, and anything it didn't find, I had the USB stick in the USB2.0 ports.

Had exactly zero problems upgrading. Plus, I had to reinstall exactly zero software. And of course the mildly OC'd Ryzen 7 1700 absolutely destroys the quad core Xeon (even overclocked to 4.2ghz).

 

[gn842a]

That sounds like the kind of thing I might better be able to do if I built two or three desktops a week rather than once every four or five years. But that's all about how to update the current hardware.

As it happens I think you are correct that the problem is overheating. I have speedfan installed but it does not seem to do a good job of thermal management. If I remember to do the settings on the fans manually upon setting up, I don't seem to have these issues. I'm wondering if there's a better product that would set my defaults upon boot for the fans, speedfan always seems to reset.

Experimentation did produce one very interesting result. Running the box fans at 100% and the CPU heatsink and GPU fans at 50% is *better* at thermal management than running the CPU and GPU fans at 100% and the box fans at 50%. As a noise control issue, the CPU and GPU fans are the noisiest so this is something of a big plus to discover. I suppose it's obvious to experienced users but it's news to me. When gaming the system runs at a relatively cool 53C with CPU and GPU at 50% and the box fans at 100%. Turned the other way, box fans at 50% and GPU and CPU at 100%, temps can rise into the 60s (and higher). When not gaming, it really doesn't matter what I do, CPU will be down around 30. It's at 29 as I write.