Ongoing *VERY* weird issue with P370SM3. Any thoughts welcome.

Ok. I'm going to write a book on this. I have a lot of information about this issue, but I still have not arrived at a cause. I am getting frustrated with this because of soon-to-be-seen obvious reasons, but what I want to know is a potential fix and cause of the problem.

Basically, I am experiencing a "shutdown" issue. This is to say that the machine will simply shut itself off. It does NOT black-screen power off, nor does it particularly "crash". If in windows, I will simply, without warning, see a "shutting down" screen show up, then it forcibly turns off as if I held the power button down. This issue is somewhat random as to when it will happen. I need to give more information as to why it is "somewhat" random. Let me begin.

1 - Once I experience ONE shutdown issue in this manner, they will become incessant until I shut off the laptop and remove all sources of power for 48 hours or more.
2 - Once the first one appears, it is 100% random as to when any consequent shutdown may appear. It could take 12 hours. It could take 5 minutes. It could shut off before windows properly boots, prompting the next reboot to state that windows did not start correctly the last time. It is very difficult to save data from the machine once this begins, due to the random nature of the shutdowns.
3 - The shutdown can happen while in the BIOS menu, or when installing windows. This rules out 100% the fact that the OS is the problem. Without an OS even installed, it still exhibits this shutdown issue.
4 - The machine ALWAYS restarts after a shutdown. I do not tell it to, I do not touch the power button, it simply does.
5 - The machine can (and usually does) POWER ITSELF UP as soon as I apply power to it. This is to say, if I remove the battery and power supply, then plug in the power supply alone, it will start itself up. Don't believe me? I'll be posting videos.
6 - The shutdown issue ceases occurring after I shut down the machine for the aforementioned 48 hours and remove power sources from it, but will restart exhibiting the problem after <s>approximately three weeks</s> the problem occurred again on the 29th, marking only a 11-day difference between the last refresh. Either the problem is getting worse or there's a 10-day limit and anything after that is luck. So far the problem has happened three times and three 48-hour+ off sessions has stayed it from occurring again. The last off session ended this morning, in fact.
7 - After a few shutdowns, I believe somewhere between 7 to 10 of them, Windows starts to freeze and hang on boot. This does not change even with a system restore, and only fixes with a reinstalled OS, which is more headache to keep repeating every <s>three weeks</s> time it happens. The windows freezing issue is displayed in the videos I will list below.

Now, here are three videos which exhibit the issue. The camera used to take them was very low quality, so the videos don't look very good, but they manage to show the necessary information.

Video #1
Video #2
Video #3

Now that that is out of the way, onto the specs. They are in my sig. Installed programs are of no relevance, since it has restarted itself mid-windows install before and also within the BIOS menu before.
I use the machine with a -80mV adjustment on my CPU dynamic clock voltage and my GPUs are using svl7 and johnksss's BIOS, which is necessary for 120Hz and 3D gaming.
Please note that the shutdown issue is NOT caused by heat nor by any overclocks, as I've not had a need to OC this machine... well ever.
Below are steps which I've taken to correct the problem, which have all failed:

1 - Removing all peripherals, battery and PSU for 10 hours.
2 - Reinstalling Windows. The windows version I used is Windows 8 Pro x64 from Microsoft's own Windows Setup Tool as I have a legitimate Windows 8 key. It was installed via USB with a Markvision USB drive, then upgraded to Windows 8.1 later. It was during a Windows 8 installation which caused the problem I had.
3 - Reinstalling my machine's various drivers, before I realized it was a problem that was not dependent on the OS.
4 - Going into the BIOS and selecting "Restore defaults" then saving and restarting.
5 - Booting the machine without peripherals, purely on battery.
6 - Removing the two RAM sticks under the motherboard (suggested by Mythlogic) effectively leaving me with only 8GB of RAM.
7 - All four original RAM sticks that came with my machine were replaced by Mythlogic when I sent it in for RMA. I had Patriot memory when the problem began, I now have Crucial/Micron and it still occurs.
8 - Manually clear CMOS battery (tried 29th June 2014 after the problem happened again).

I was initally of the opinion that it was a power supply fault, but the issue happens
A: On battery with no power supply plugged in.
B: With both battery and power supply plugged in.
C: On power supply without battery plugged in.

Which effectively removes both the battery and the PSU from the equation.

What I have noticed is that my -80mV remains active after 6+ hours of no power supply devices attached to the PC, however after the 48+ hours disconnection, it reverts to +/- 0mV. <s>This leads me to believe that the 48+ hour disconnect is akin to a CMOS reset. I do not know where the CMOS is for my machine, so I have never tested the CMOS reset manually, however since defaulting the BIOS does not fix the problem, I believe it's most likely a board issue.</s> EDIT: Got problem again; tried CMOS reset. Not only did my undervolt vanish, but it changed the UEFI boot mode, which never happens after the 48 hours, so I can confirm that 48 hours does NOT evoke a CMOS clear, and thus this is not a potential fix. It also took me 15 minutes to reapply -80mV and UEFI to boot Windows 8 instead of "other" because it kept shutting down and restarting while in the BIOS menu.

Now, I want as much input from others as to what the issue could possibly be. I am so far believing it to be somehow a bad BIOS install or just a bad board. This problem started showing up at the end of March 2014, and I have been using the machine since the end of September 2013. Since it originally showed up, I've not been able to rid it. I once sent it up to Mythlogic in April, and they checked it out but could not find/reproduce the issue, saying that my hardware all checked out. As it was off for over 48 hours prior to sending it up to them though, which I've now noticed to be the cause for it to work well for a few weeks, the problem would not have shown up unless they kept and used it for about a whole month.

If there is anything else I can provide information-wise, please let me know. I am just beyond frustrated as I've had disjointed use of the machine for the last two months and I didn't even have it usable for the entire of April, and though most of the start of April was indeed my issue where I had delays in being able to ship it out, the fact is that it can literally happen at just about any time, with my doing anything, and it can potentially cause me serious problems if in the middle of a project or something when it decides to pop up. I really need to fix it. I just want my laptop to work man, without all the stress.

EDIT: Until I get the board replacement, I'll keep this thread updated, just in case I'm the unluckiest person on this planet and it happens again with the new board. Instead of deleting information, I'm going to <s>strikethrough</s> anything that is proven false by repeated experiences of the issue, so you can see how my information appears to fall into place.

 

First of all, sorry you had to go through all that, it sounds extremely frustrating indeed. I'm definitely grasping at straws here, but have you tried using only 1 ram in 1 slot at a time just to rule out the possibility of a faulty sodimm slot? I vaguely remember seeing similar symptoms like yours and it was traced down to a faulty sodimm slot on the motherboard.

Also, might want to contact Mythlogic again and send them those 3 videos, so they at least know it's a real and reproducible issue.

 

I have contacted them and shown them this, and they have essentially acted like it's still a software issue, which is leading to my increased frustation. I can possibly try 1 RAM slot at a time, but see here's the thing. In order to troubleshoot this, I must keep my machine in an un-usable state. I cannot do more tests until I see the symptoms appear again, and I am not keen on not having a computer to use really. The only other machine I have in a usable state, as my D900F simply flat out refuses to even boot to BIOS, is an old HP scraptop that my friend put together from stuff they were literally throwing away out of his job. It's got a battery, but no wireless card and it's an AMD turion x2 CPU with 2GB of RAM. I can't even watch twitch livestreams or play terraria on it, really. And that's the only machine I have to use since my desktop just up and died about a month after I got this machine, and we haven't had the money to replace that yet. Since my last 48 hour off cycle ended today, it likely won't show up again for at least a couple weeks. I also can't simply get my machine up to Mythlogic in under 48 hours (as it needs to be off with no battery in it to ship) so they can experience it firsthand.

 

Hum, maybe send Mythlogic a PM on here and point them to this thread? I kinda expected better service than that tbh

Since your 48 hour cycle ended today, why not try the ram thing starting today? Remove all but 1 stick and run it for a few weeks. If you don't get the problem again after say 5 weeks, then perhaps it really was a faulty ram slot. (although I guess this is contingent upon whether that 3 weeks estimate you gave actually holds up every time you do the 48 hour cycle)

I mean honestly at this point, this is pretty much the only thing I can think of off the top of my head, since you've already ruled everything else out (and Mythlogic says your hardware checked out OK). Although, the weirdest part is the problem disappearing for a few weeks after the 48 hour cycle, then coming back again.

 

Yeah. That's the truly weird part. It's that it vanishes after the 48 hour cycle. I didn't notice the first time I sent the hardware up, you see. I can confirm that 10 hours off does not help, and that 24 hours off does not help. I am not sure exactly HOW long it takes, but I am fairly certain that 48 hours or so is the time limit. I was even thinking maybe it was a BIOS issue conflicting with windows 8.1 in some sort of bit rot, and the essential forced CMOS reset from leaving it off that long just kind of reset the problem timer. But I have no way of knowing, and no way to prove this theory. It's a real brain wracker. I also would be limited to 4GB of RAM if I tried that 1 slot thing from now, and I can easily use up 8-9GB normally. If it happens again anytime soon, I'll try it. I already confirmed that the two slots under the keyboard not having RAM in them do NOT affect problem occurring, so I will probably take out the RAM under the motherboard and only use those under the keyboard and see if it still occurs. There's nothing else I can think to do.

 

Hmmm, next time it happens try removing the CMOS battery and doing a power drain and see if that fixes the problem right away instead of having to wait 48 hours.

 

Do you know where the CMOS battery is located? If that fixes it for sure, then that means it's something BIOS-level... though I don't know what. It frustrates me equally that I can't figure out and fix this issue just as much as it does that it causes usage problems.

 

Should be under the keyboard, to the right of the lower ram slot. It's got a black plastic cover over it so it won't seem obvious on first glance.

P.S. Unrelated, but you might want to check and see if your master 780M has a slight warp in it like mine. I'm thinking it must have deformed due to the mechanical stress placed on it, since one of the corners of the card rests on a hard, plastic square "block" on the motherboard. (it's to the left of the mSATA screw, you'll know what I'm talking about as soon as you open up the back)

 

I think mine is properly flat, and I have no real heat issues with it, but I can check that later. Aside from this issue, my machine is essentially all I could ever ask for.

 

What are your auxiliary temperature readings like? (South bridge, hdd etc)

 

This is it running idle, or how it would be from a cold boot, etc. https://hostr.co/file/BfgpPzs69AoW/Screenshot1027.jpg

I don't know if there's a temp you need to check that isn't in that HWMonitor cap, but I've most often gotten shutdowns without me actually doing anything demanding (like watching a video or just typing something up). I do not do any repastes or anything of that manner when I leave it off; I simply disconnect it and turn it back on after 2 days. If it magically overheats at those times, then it doesn't overheat any more after the 2 days of being off.

 

I prefer HWInfo since it gives a lot more info, and reports the PCH temp as well, which HWMonitor doesn't.

 

Intel PCH is approximately 64 degrees. No idea if that's good or bad, as I've never seen that temp before XD.

 

Pretty decent considering how hot your room is. So yeah that's not the issue. Really hoping it's as simple as a faulty ram slot on the motherboard.

 

Yeah. Well I got Mythlogic to agree to a board replacement, which I am about 555,000,000% sure is the problem. Now I just need to send it up again. But I still want to know what the hell the problem is.

 

PCH is safe up to 100c.

 

Well, any other ideas about this Meaker? With a board replacement I suspect I'll have the problem go away, but I still want to know what it is. It's my curiosity that's killing me.

 

I would try booting with usb live linux system and trying to work for an hour or so to rule out hardware problem.

 

If it happens again that's a possible thing I could try, but once more, I have gotten shutdowns while I was in the BIOS. That should be way before any OS has a chance to do anything. Also, I'd need to use it for over 24 hours in linux to be certain. After I took out the two RAM sticks that I listed I did, it took 12 hours before a shutdown, but afterward it turned off rather frequently, within 2 minutes of rebooting. Once the issue has started however, it hasn't lasted up to 24 hours without a shutdown, even if it lasts 12+ hours.

 

Dodgy solder job on a board in the machine perhaps then.

 

That's why I couldn't figure anything out. The whole issue was so weird.

 

Ammended first post. Tried CMOS reset. Did not help. Also, the problem occurred after 11 days this time, and thus is not limited to 3 weeks. Unless the problem is simply getting worse; I cannot say

 

I don't think there's any other reasonable explanation at this point. I hope you get a new board soon.

 

I'm aiming to send the PC up hopefully early in the week starting July 7th. Won't have the money before then. I really really want that new board. I hope I don't go without my PC for another month again. I'm trying to be regular with my livestreaming too and it's really putting a damper on it when I keep missing my PC for days and weeks at a time in random intervals, far less when life gets in the way naturally. All I know is I just hope things work out soon. I NEED this back working without issues before September when school re-opens for me.

 

How bad is shipping from Trinidad to continental US? I'm thinking it would be at least $150 USD since the only option is to ship by air.

 

It is ~$150 USD exact, actually. UPS next-day shipping insured. But to collect it on the return trip using a skybox it's a further ~$50 USD (shipping to skybox is free due to free USA shipping under my warranty). Just that $1 USD = $6.45 TTD after conversion. So it's quite expensive. We don't make a massive amount of money to compensate; minimum wage here only translates to $1.96 USD/hr or so.

 

Gonna test 27 hours of off time later today and see if that fixes it. I'm hopefully sending it up next week so might as well, right?

 

Actually hang on, they charge shipping for warranty returns? But I guess it's because you're in Trinidad so shipping isn't covered.

 

Yea, it's because I'm in Trinidad; the international shipping isn't covered.

 

Welp, 24 hours appears (so far) to have done the trick. But just over 10 hours doesn't. Odd indeed.

EDIT: Confirmed, 24 hours off does remove the bug. Uptime of 1 day 11 hours at time of writing. Not sure how long it'll last though, of course.