TR Forums

Thought I'd think out loud for suggestions while I think about how to phrase a believable RMA request.

Machine is an i5-6500 / H170, 16GB(2x8) DDR4, RX 460, blah blah. I have isolated the problem down to either the motherboard or CPU. Changing the PSU or GPU (or yanking the GPU entirely) has no effect, using a completely fresh install of Windows 10 has no effect. Clearing CMOS has no effect. Memtest86 V7.3 has yet to find a problem with RAM (given the how quickly and often the crashing happens, I would expect it to easily find something if RAM was involved). I have no equivalent spare RAM, CPU, or motherboards to swap out for testing. All temperatures seem normal. Problem has manifested last Tuesday after user had it for four months, has been fine up until that point. User was running the machine in a closed cupboard though, so they may have cooked it.

Here's what's happening: shortly after booting into Windows 10, it will crash. Timing is random, but while playing with it crashing didn't seem to take longer than 10 minutes. Using it "normally", it doesn't seem to make a difference what you're doing, eventually it will BSOD, and there are so many different bugcheck codes that it's not even worth mentioning them. It does seem to be able to boot to desktop without any trouble. Before it crashes, it may or may not choke on whatever it's doing (e.g. fails to read or run something). BUT - if I start Prime95 in test mode fast enough, it becomes relatively stable. Not *completely* stable, it will still fumble tasks/writes/whatever every now and then, but it no longer seems to BSOD (Prime95 itself runs mostly OK, though it has detected one rounding error so far). Loading the RX 460 instead (i.e. generating a load on the PSU without loading the CPU much) does not have the same effect, it will BSOD as usual.

You may think, aha, this is a C6/C7 PSU underloading problem. I don't think so - as I said, I have tried two different PSUs that are stated to be "Haswell compatible" by the manufacturers. Motherboard UEFI has an option for a dummy load (irritatingly, not for C6/C7 - I suspect that those C-states are not even enabled at any point).

Does it seem reasonable to think it's a bad motherboard? Perhaps it cooked a VRM?

Ugh... hate problems like this. I'm gonna go with RAM, but that's just a wild-ass guess. Multiple sticks? If so, try removing sticks one by one to see if it stabilizes.

I assume you've already done scans with multiple malware detection/removal tools?

Have you tried Windows Memory Diagnostic?

Have you ruled out a disk drive going bad?

I had a laptop that would regularly crash, but pass a Memtest86 scan. The problem was bad memory. To find it, I had to boot into Memtest86 immediately after the system crashed, and Memtest86 would find errors. If I waited a minute after the crash, then run the memcheck, no problems were found. No crashes after I replaced the memory.

FWIW, Memtest86 has since had a couple of single bit errors, but IDK if that's actually the RAM or what. I'm not done with letting it run yet, will see when I have more data

LostCat wrote:

Have you tried Windows Memory Diagnostic?

Have you ruled out a disk drive going bad?

Yes (in addition to Memtest86) though WDM found nothing. And yes, have swapped drives, no effect

just brew it! wrote:

Ugh... hate problems like this. I'm gonna go with RAM, but that's just a wild-ass guess. Multiple sticks? If so, try removing sticks one by one to see if it stabilizes.

I assume you've already done scans with multiple malware detection/removal tools?

Swapped drives and ran from a totally clean install, no effect

Can PSUs fail in a way where at low loads they freak out? Is that even possible?

derFunkenstein wrote:

Can PSUs fail in a way where at low loads they freak out? Is that even possible?

I was wondering about that before I eliminated PSUs. My electrical knowledge ain't good enough to have an answer. There is the whole shebang about Haswell power supply compatibility, of course

try to run the gpu and the cpu at max power states, disable core parking and speedstep (or the name it has now). is the rx460 connected to the psu or just to the pci e for power??

If you get ANY errors with Memtest, that blacklists the RAM, so start by RMA'ing the memory and let them know that Memtest86 generated signle-bit errors. Othewrise, motherboard (CPU is highly unlikely). Rule of thumb for RMAs is that the person processing the request may not be a serious techie, they'll just read from the playbook. IOW don't try to explain "it only happens when the system is idling." For memory, errors in Memtest should be sufficient. If it's the motherboard, let the company know that you are not overclocking, you tried updating the BIOS, and you reinstalled Windows (you did all those, yes?), and that the crashes continue. Even better if you pulled the GPU and tried using the integrated CPU graphics, proving that a GPU problem is unlikely.

TwistedKestrel wrote:

FWIW, Memtest86 has since had a couple of single bit errors, but IDK if that's actually the RAM or what. I'm not done with letting it run yet, will see when I have more data

There ya go, bad RAM or bad RAM settings. That Prime95 also had rounding errors pretty much confirms it.

Kougar wrote:

TwistedKestrel wrote:

FWIW, Memtest86 has since had a couple of single bit errors, but IDK if that's actually the RAM or what. I'm not done with letting it run yet, will see when I have more data

There ya go, bad RAM or bad RAM settings. That Prime95 also had rounding errors pretty much confirms it.

I don't think that does confirm it. You run Memtest86 long enough on anything and you will get errors. RAM can of course fail in different ways, but if it were bad enough to cause the crashing I was seeing I would expect to see a lot of errors, and to see them often.

I think the errors that Memtest was seeing were not memory errors at all, but just an artifact of whatever is actually going on. I'm RMAing the motherboard, so we'll see

TwistedKestrel wrote:

I don't think that does confirm it. You run Memtest86 long enough on anything and you will get errors. RAM can of course fail in different ways, but if it were bad enough to cause the crashing I was seeing I would expect to see a lot of errors, and to see them often.

I think the errors that Memtest was seeing were not memory errors at all, but just an artifact of whatever is actually going on. I'm RMAing the motherboard, so we'll see

I suppose it's possible. But Prime95 blend runs tend to be a better tester for RAM stability. Have you checked the administrative event logs for WHEA hardware errors?

RAM issues are not simple cut and dry affairs with lots of crashing. You can have unstable RAM that crashes a desktop less than once a month, or every two minutes. It's a factor of just how unstable it is, as well as what tries to access the unstable portion, or if the unstable portion is ever even used as is the case with 32GB and 64GB kits.

Turns out it was the CPU and the RAM. Also I don't remember RMA'ing things being such a nightmare, I could swear it used to be somewhat common to have warranty services in Canada.

THAT'S EASY! COIN-OF-THE-WEEK MINING FOR STABILITY!

Kougar wrote:

TwistedKestrel wrote:

FWIW, Memtest86 has since had a couple of single bit errors, but IDK if that's actually the RAM or what. I'm not done with letting it run yet, will see when I have more data

There ya go, bad RAM or bad RAM settings. That Prime95 also had rounding errors pretty much confirms it.

True statement. I installed a kit of RAM in my machine a couple years ago (?) and didn't have any issues until TR Frankenbot F@H in October. IIRC, I couldn't even keep my client running for more than 20 minutes before it'd crash. I run Memtest and/or Prime 95 on all my new kits, but usually only for 12 hours. This kit didn't spit errors until ~ hour 19-21. RMA'd the RAM and the problem was fixed.

What brand is your RAM? G.Skill has a warranty/RMA that's very easy to work with.

The RAM is GeIL. Their RMA process was eh. Kinda similar to Gigabyte's RMA, except they communicate even less but they ship faster (once they tested my RAM, they shipped out a new set from Taiwan and I got it the NEXT DAY). I ran Memtest86 on the replacement set for 21 hours with the *bad* CPU and got no errors.

TwistedKestrel wrote:

The RAM is GeIL. Their RMA process was eh. Kinda similar to Gigabyte's RMA, except they communicate even less but they ship faster (once they tested my RAM, they shipped out a new set from Taiwan and I got it the NEXT DAY). I ran Memtest86 on the replacement set for 21 hours with the *bad* CPU and got no errors.

Just a note that 21 hours may not be enough, but the number of passes. I once had a dual channel kit (granted it was only 2x1 or 2x2?) failed after its 13th pass, which took >16 hours. Now that we have like 16 or even 32 gigs, this madness is only going to take longer. Good thing memtest finally has multicore mode and can go faster.

Looks like this system is coming back to me now with the exact same problem. The machine had been good in the interim, but I don't know what to make of this. Clearly not a randomly defective part...

Can you disable Intel SpeedStep?

If the RAM RMA fixed it before, who's to say it's not the same thing again. Crazier things have happened after a year and a half of usage.

It is reverse of expectations.

I am going to go with a bad core CPU power supply. It is not regulating well under light loads, likely due to cooked capacitors.

Otherwise, memory.

Could be CPU. Unlikely but you've ruled out the RAM as best as you can by getting an RMA and trying different RAM.

Identical problem on different modules means that it's incredilbly unlikely to be the RAM at this point.

Don't forget that RAM errors in Memtest can be a bad memory controller on the CPU or a slightly bent pin in the motherboard socket, too. Since failing memory controllers are something I've seen multiple times on Haswell chips with the FIVR, and NEVER on any other Intel CPU, I'm going to call out the motherboard. Can't be 100% certain but if you have simultaneously tested the new RAM using a different PSU and using the IGP to rule out the GPU, you're safe to RMA the board and explain the reasoning. I'd probably lie and say that I'd tested a different CPU in the board with the same results just to tick that box on the RMA monkey's triage form.

I'll try not to post too much until I actually have the thing in my hands, but I should clarify:

I can't say with 100% confidence that the RAM was actually defective last time, precisely due to what Chrispy said. The RAM got replaced *before* the CPU. And in the process the motherboard got RMA'ed as well. (I had no intention of needlessly requesting replacements, but I did do this through proper channels and they could well have sent the originals back)

TwistedKestrel wrote:

I'll try not to post too much until I actually have the thing in my hands, but I should clarify:

I can't say with 100% confidence that the RAM was actually defective last time, precisely due to what Chrispy said. The RAM got replaced *before* the CPU. And in the process the motherboard got RMA'ed as well. (I had no intention of needlessly requesting replacements, but I did do this through proper channels and they could well have sent the originals back)

I hope you figure out the problem soon. After enough RMA's you might end up with a PC of Theseus situation...

It's back in my hands, and it's exactly the same as before. I re-seated the CPU for good measure (and looked for bent pins / scorched pads), tried disabling Speed Step... nothing I did seemed to make a difference at any point. I expect to be starting another RMA with Intel, because I'm not sure what else to do

100% reproducible instant crash - run Prime 95 in max power consumption mode for a few minutes, then hit stop.

Concupiscence wrote:

I hope you figure out the problem soon. After enough RMA's you might end up with a PC of Theseus situation...

That ship has already sailed... the RX 460 that was originally installed was swapped out by the user, wound up back in my hands, and I ended up RMA'ing it a month before the warranty was out (was using it while performing surgery / waiting for an AIO kit I never ended up using on my HD 7950)

Hmmm, just found an interesting thread on the Intel forums...

https://communities.intel.com/thread/97358

Maybe this is too early to call, you be the judge...

The TLDR; summary of the thread above is that some Skylake CPUs have a power management related defect where they have problems entering/leaving C-states. I disabled all C-states on the machine in question and now the 100% reproducible crash no longer crashes it. I left it running overnight and it hasn't crashed yet. I couldn't tell you why this would manifest after running fine for a year, but I feel pretty confident that RMA'ing the CPU again is the correct course of action.

Problem is gone with replacement i5 6500... don't really like that "lightning" has struck in the same place twice. Oh well, it seems the warranty is tied to the physical CPU and it looks like it got reset to three years again, so I guess I'll do this again if I have to