More A10-7870K Annoyances

[EXCellR8]

This little computer is driving me crazy. It's a Gigabyte F2A88XN-Wifi 3.0 (latest BIOS 'F6') with a A10-7870k and 4GB DDR3 memory. I've been using it with Ubuntu 16.04 for quite some time, and for the most part it's fine, but every now and then it seems like the video freezes and I have to hard reset. I just chalk it up to buggy proprietary drivers and go about my business--as it doesn't happen too often.

Fast forward to last night... I split the boot drive in two and install Windows 7 on one half. Everything installs fine but once the graphics driver is installed I experience the same issue as on the other OS. The screen freezes up and I see colored artifacts throughout the display; there's nothing that can be done to fix it aside from a hard reboot. Temps are fine and BIOS settings don't seem to affect the problem whatsoever.

I have the same hardware in another machine running Windows 10 and I've never had an issue with it, so do you think the chip in this one is faulty? I can replace it easily, and probably should, but I figured I'd get some input on this first. Thanks in advance.

 

[LTC8K6]

Swap the chips around and see if the problem moves to the other computer?

 

[EXCellR8]

I think I might try a discrete video adapter before I do that, because now that I think about it the other machine, which I use as a steam gaming server, is using a FirePro card and not the APU graphics. I'm interested in seeing if maybe that fixes the issue and if so, that works for me. If not, I'll dismantle the machine and see if it does the same thing in another board. To my understanding, these boards didn't have Godavari support out of the box and had to be flashed, so perhaps there is still some kind of APU graphics bug.

 

[VirtualLarry]

APU graphical glitches and freezes / hard-locks can also be down to bad system RAM.

 

[William Gaatjes]

Maybe try some overnight testing with memtest ?

 

[amd6502]

APU graphical glitches and freezes / hard-locks can also be down to bad system RAM.

That's my guess. OP should definitely run MemTest86 or a good ram check util.

MSI and Gigabyte have been rock solid in linux fm2 APU stability (I had one low end Asus A55 where I had to replace an A8 with an Athlon x4---the Asus board had bios firmware load to video region of linux gpu driver and cause consistent instability).

Even if a discrete gpu fixes the problem, the memory should be tested for 100% reliability.

As an aside, 4mb is pretty little RAM, so either an ram upgrade, or a discrete gpu to free up a little ram would be good.

 

[EXCellR8]

Thanks for the input, good sirs.

I'll run memtest overnight tonight but the discrete GPU (and disabling APU graphics) has not produced the same result so far.

I used a very unremarkable HD4350 low profile card and it's working just fine at the moment--but will continue to monitor it.

 

[VirtualLarry]

I have occasion "Driver Reset" issues with the video driver on my Lenovo laptop with a quad-core small-core AMD APU.

 

[EXCellR8]

Ran memtest overnight but it was fine; I left it on before leaving for work this morning though.

I have some other DIMMs I can try also but I don't think it's the memory--but not ruling it out. The APU seems to be much snappier with the R7 component disabled.

 

[amd6502]

well, memtest86 is really thorough so it looks like this just about rules out bad memory (there is a very remote possibility I can think of that the RAM might only act up at higher case temps). So I am totally stumped what this could be.

Could it be a bad APU? maybe (and worth testing it maybe if you have a spare board). also maybe likelier it could be an aging mobo? How many years old is the board? (I did once have one aging board I picked up from a thrift store that had a variety of issues.)

Other old silicon (on the mobo, or APU) can also act up under higher temps. I'd try to make sure the case cooling is decent, especially if it's a small form factor build.

Drivers seems unlikely because you tested under multiple OS' (win and linux) and even have a working machine with same specs. Perhaps try to use similar BIOS settings, or reset to bios defaults? Also, maybe it could help aging silicon to slightly overvolt various parts, like southbridge (mobo volts) or northbridge (APU mem controller voltage). Or you could go easier on the VRM transistors by running under a 65W cTDP.

If the computer is sitting very close to a router could there be a remote possibility of RF noise causing issues? (they do make some crazy very high wattage routers these days).

 

[EXCellR8]

Not sure but I suppose the easiest and quickest way to further troubleshoot would be to simply replace the APU, which would certainly indicate if it was the board or not. That might actually be the issue though because there are a few other anomalous things that have happened, like: seemingly random boot failures, video loss at POST, and the occasional hard lock. Perhaps the board is getting too hot and some component is just having a hard time with heat. There isn't a lot of ventilation in there either so that's completely possible, though PC health in BIOS doesn't indicate any voltages or temps out of the norm. I did notice that the system clock never seems to be right either...

A driver issue probably wouldn't be consistent across different OS so there's gotta be some underlying hardware fault... just gotta find it. I'll take the whole thing apart and bench it this weekend if I can find a little time.

 

[ZGR]

I would reseat the DDR3.

 

[EXCellR8]

Ugh this little computer is such a turd... swapped CPU and the problem is still there, but worse now so I'm replacing the RAM--even though MT didn't find any errors. If that doesn't fix it then I guess it's a lost cause and I'll build something else. I really have no idea why it's still acting up but BIOS settings don't seem to affect the stability/freeze-ups. Unreal...

 

[EXCellR8]

Seemed to be okay with slower DIMMs but it's still doing all sorts of wonky nonsense. I think the board itself might be on the way out or something.

 

[EXCellR8]

Okay, so I'm 99% sure that this issue has finally been resolved... so I figured I'd post my ultimatum.

I believe it boiled down to a unstable DIMM and a buggy graphics driver of all things; less to do with the actual APU itself. I was benching the system with the original DDR3 1600 sticks and I was able to run it with a single 2GB but not the other in the same slot. One of the sticks caused the computer to lock up, as did both sticks when installed. I swapped both modules for 2x2GB 1333 DIMMs and the system ran fine for a few hours (the problems I experienced typically happened within 5 to 10 minutes). In the mean time I had ordered a pair of 2x4GB low voltage 1600 modules.

Both sticks of the DDR3L are installed and I have yet to encounter any issues there. I removed the AMD drivers and reinstalled the chipset drivers that I downloaded from the board manufacturer's site; not from AMD directly. So far, the machine has been running perfect so I'm crossing my fingers that I've figured it out. I also tested the SFX power supply (a Seasonic) and monitored the CPU temps while the computer was on the bench. These A10's do run on the toasty side but the cooler is adequate.

In any regard, thanks for the assistance