Random PC Freezing - no pattern

I'm actually not of the opinion that it might be the PSU (on both ASUS and M910t machines???) which is suspect.

The ASUS machine has a 600W NoFan PSU but in fact the machine itself uses around 220W total even with both monitors powered on (everything is running through an APC UPS, so I know how much is being covered by battery backup). The new M910t only has the stock 250W PSU from Lenovo.

The GTX 1050ti is rated <75W TDP, but in fact I don't "game" and it typically runs at under 40W max.

I honestly don't think I have either inadequate power (on the M910t) nor flaky power (on either machine).

But this afternoon I did download five legacy NVidia driver versions going back to early 2018 when I know I had none of this instability. There's really nothing that's been added in newer driver versions which impacts the older GTX 1050ti card as I use it, and again I'm not a gamer so am non-impacted by any game-related fixes in newer driver version.

If the latest Aida64 tweaks don't seem to help I next plan to downgrade my installed NVidia driver, in order to see if my suspicion might be valid about a newer NVidia driver version perhaps introducing some issue. I definitely seem be "much more stable" (but not yet back to 100%) having turned off screen saver and power-save mode.

Anyway, current clock (since last freeze/re-boot) on the ASUS machine is at 1 day 6 hours, and the M910t shows almost 22 hours. Been able to watch/record TV, work, remotely connect with RealVNC and Team Viewer, etc., all apparently without mishap. We'll see how long the current streak remains in effect. Naively for now it does seem that turning off all of those "lockup might occur in rare cases" hardware monitoring settings in Aida64 has brought about an improvement.

Yeah, I thought it would be a leap to have problems with PSUs on more than one system but just thought I'd throw it out there as a possibility.

If you read the release notes for driver updates at NVIDIA, most of them are to accomodate gamers. They are the ones driving the desktop business these days.

Greybear, in that forum, seems to be very knowledgeable about these kinds of problems. I read some other threads in which he was trying to help and he knew where to look for the cause and was able to provide details about what to try. Very technical. Perhaps you might want to ask about it there.

Well, I'm setting new records for recent stability duration thanks to disabling those lower-level I/O sensor readouts from Aida64: M910t has now been up for 1 day 10 hours, and the ASUS has now been up for 1 day 17 hours. Both continue to be fully operational including still being connectable through RealVNC and Team Viewer.

I've now downloaded older versions of Aida64 to try again (with those I/O sensor settings turned back on), since I know this issue is only about 3-4 months old for me.

Current "stable" Aida64 - 5.99.4900 dated 11/10/2018 is the one I'm currently running on both machines (and that I've disabled those low-level items to avoid "may cause lockout in rare situations").

I now also now have retrieved older Aida64 5.98 from 09/20/2018, and Aida64 5.97 from 03/28/2018, and Aida64 5.95 from 11/27/2017. I know I must have been running with all of these over the past year with the low-level options active but also without the instability I'm currently observing, with the same ASUS machines (including the one which is now dead and gone). These older legacy versions are also running on other Lenovo desktop machines so I'm confident I can just use them without concern on my M910t.

This is encouraging. In addition to newer NVidia drivers installed over the past few months, there were these 5.98 and 5.99 versions of Aida64 installed around 9/20 and 11/19 respectively. And based on what I now see using 5.99 but without its low-level readout options active, maybe I'm really on the right trail now.

We shall see.

Well, I had a dentist appointment today. When I returned late afternoon the ASUS machine had frozen at 4:50PM. So it made it 2 days 2 hours before giving out. The M910t is up and still running fine, 1 day and 18 hours since its last re-boot.

Just to be proactive I decided to back-level the ASUS machine for both NVidia graphics driver as well as Aida64. It seems that turning off the low-level sensor readouts in Aida64 had terrific effect on both machines (e.g. the M910t is still ticking since I made that change), but since both of these software products seem relevant and since I never had instability until installing recent updates to both late in 2018, I decided just to roll back both of them to get me back to a stabler environment. There is zero new or fixed functionality that, for me, justified applying any updates to either software product over the past year. So reverting to what was running perfectly fine a year ago is of most concern to me.

So I've now reverted to Aida64 5.95.4500 which was dated 11/26/2017. This replaces 5.99.4900 released 11/20/2018. I also have re-enabled all but one (SMBus operations) of the items I had been running previously so I once again have all of the hardware monitoring values I had always been displaying. The SMBus item doesn't impact what I do monitor, so turning it off is of no consequence.

And I've now reverted to NVidia WHQL 391.35 which was dated 3/27/2018. This replaces 417.71 which was released 1/15/2019.

Let's see how things go. I will defer possibly rolling back anything on the M910t until/unless it next freezes.

One more note regarding Aida64, and my ASUS motherboard. Late last year (and this may have been tied to the software update, rather than to my motherboard) I decided to turn off monitoring the temperatures of my three internal drives as reported by SATA/SMART or the PCIe/NVMe. Every so often I would observe that the temperature items for the SATA drives would simply disappear from Aida64. At the same time Windows Explorer (and Free Commander XE) would struggle to display the drives. I was unable to determine or decide if I was seeing a problem with Intel RST (which is installed, presumably to improve SATA performance for drives running AHCI, not RAID) or with the SATA controllers on my ASUS motherboard.

To avoid possible contribution to the issue which might be flaky SATA controllers, I decided to stop monitoring drive temperatures in Aida64. I also un-checked the related "SMART operations" setting, which was one of those "stability" group of settings. So I've been running with that option disabled and no drive temperature reported for a good number of months. For the time being, even though I've gone back to Aida64 from 11/26/2017, I'm not going to re-enable SMART functions and monitoring drive temperatures. I just want to be able to see what happens with the rollback of both Aida64 and NVidia.

We shall see.

Why fool with Aida64? Just uninstall it to see what happens. It's not really necessary.

I think Firefox may still be causing your random freezes. I suggest not running it for several days. Since you're not around when these freezes occur, you wouldn't see the graphics driver stopped and then recovered error. This event occurs and recovers within about 30 seconds to a minute. Happened again to me yesterday when I was looking at Videos at YouTube even though it had been running for a week without a problem.

dsperber said:

this very setup has been working 100% reliably since 2010

You have very old hardware. Heat or oxidation kills all hardware eventually. But before death there is usually a frustrating period of intermittent failures that are assumed to be caused by software or setup when in fact they're caused by failing hardware. my two cents...

johnhoh said:

You have very old hardware. Heat or oxidation kills all hardware eventually. But before death there is usually a frustrating period of intermittent failures that are assumed to be caused by software or setup when in fact they're caused by failing hardware. my two cents...

When I said I had been running with this setup since 2010, I didn't mean I haven't upgraded hardware in those past 9 years. I simply meant I have been running the very same notion of WMC HTPC and essesntially the identical software for this period.

But in actuality, I have upgraded my physical machines three times over the past 9 years. All are homebuilt, the first with a SuperMicro motherboard and the last two with ASUS motherboards. It's this most recent machine upgrade which was only built early in 2017 which is causing my headaches now. It's a Skylake Z170-Deluxe with i7-6700 CPU. Essentially it will be the final Win7-supported hardware. But essentially it's brand new hardware, though clearly exhibiting definite signs of hardware instability both in SATA and USB.

I've also upgrade through many graphics cards in the past 9 years (ending up with the current GTX 1050ti) along with several generations of TV tuner cards (both OTA/ATSC and cablecard-enabled), first from AMD and then from Ceton and Hauppauge.


But back to the case...

Had freezes on both machines again this morning (including the M910t after remaining up for 2 days and 7 hours), one occurring while I was on a conference call and looking at some emails on the screen in front of me. I went to select another email to view and I was frozen. Turns out it had occurred about 40 minutes earlier but I hadn't noticed.

Not surprisingly (anymore) I went upstairs to check the ASUS machine and sure enough it too was also frozen, sometime after 6AM this morning after being up only 8 hours. And this is the machine which was running with the back-level NVidia driver and back-level Aida64. So this is a real disappointment.

I then tried to re-boot the M910t but it froze at the Windows welcome screen! I went upstairs and sure enough the ASUS machine was frozen again, though I had rebooted it less than an hour earlier.

Clearly there is something "sympathetic" going on between the two Win7 desktop machines on my LAN. Seems to be totally immune to Win10, which is the laptop I'm using to type this post right now. Given that my M910t is literally brand new I have to believe the source of the problem is fundamentally with the ASUS Z170 machine. Interestingly both are Skylake machines with i7-6700 CPU.

I've now powered down the ASUS Z170, and also flipped the power switch on the PSU so that it should not be capable of spewing out anything over the LAN that can be seen by the M910t. Although the Z170 was my primary HTPC (with 6-tuner Ceton and 4-tuner Hauppauge) I have relegated my HTPC chores to the new M910t (with older 4-tuner Ceton and 2-tuner Hauppauge) for the past few weeks. So I don't actually need to use the Z170 in my office. I can use the Win10 laptop which shares external mouse, keyboard and Eizo monitors and have the exact same daily "work" capability I require. This Win10 laptop has remained up and available and on the same LAN throughout all of this nonsense with my two Win7 machines, and appears completely unfazed by what's going on in the Win7 network environment.

I now want to see if the brand new M910t HTPC can run all by itself isolated from the ASUS Z170 which is now truly offline indefinitely, and exhibit stability. Or is it, too, strangely subject to this instability from something coming from within itself, perhaps from intermittent TV tuner cards? This all seems so improbable.

Anyway, we'll see if I can establish normalcy with just the M910t running. I will also back-level its Nvidia drivers and Aida64, just to try and help out for this next test.

I obviously am still procrastinating replacing the ASUS Z170, either with the second M910t that I also bought or just with a replacement Z170-Deluxe motherboard. I really have to establish proof that it is a hardware issue on this machine which is at fault before I embark on either of these extreme remedy attempts.

If you want to establish it is a hardware issue, then why run aida64? Still can't ignore the Firefox issue.

wither 2 said:

If you want to establish it is a hardware issue, then why run aida64? Still can't ignore the Firefox issue.

As I've mentioned before, I remotely maintain about 23 computers for friends and family, almost all of which are machines that I ordered/configured and installed for them. And NONE of them exhibit any of this freeze symptom. Typically they can remain up for weeks or even a month or more between re-boots, which are from me doing a monthly Windows Update or other software/driver update that deserves a reboot. And every single one of those machines runs Aida64 without an issue, and typically the latest version... all without any instability.

So I can't bear to think I shouldn't be allowed to run software that I myself have been running for years, on my own machines as well as others'. But I understand why should I tempt fate at all and run Aida64 no matter how it is disabled, when by just not running it I could certainly point the finger at some other cause (like hardware) if I still see a freeze. I'm just frustrated, and can't bear to not see the Aida64 OSD that I'm used to, which also happens to show me the current time and also up-time, and its non-movement is evidence of the "freeze" even though Perfmon and Clockwise and windows clock all continue to move and right up to the bitter end I can still usually access mapped network drives from the other non-frozen machine.

Anyway, I should be able to run the new M910t "fully configured as my HTPC and with Aida64 running fully configured) in a perfect world. With the ASUS machine now truly taken down and offline completely, I'm hoping to see the M910t remain up for days and weeks, as it should.

While I wait for this proof (of either success or once again disappointing failure), I may just bite the bullet and use the opportunity to take the day or two it's going to require to replace the motherboard in the ASUS machine as long as I do have the replacement new motherboard waiting use. Worst case, if this doesn't actually solve the problem on the ASUS machine I still have the second extreme solution option of just replacing the machine completely with my second in-the-carton new M910t. But I really would like to keep the current ASUS case (and 5.25" BluRay burner optical drive which can't be transferred to the M910t which only has room for a slim optical drive).

I understand. Only reason I mentioned Aida64 is because you were concerned about the updates to it and were reverting back to previous versions. I just thought it would be easier to completely uninstall it temporarily.

You've got a lot of money tied up in all that equipment. This system is 11 years old and my other is almost 9 years old. Don't have any spare parts.