WIN 10 Meltdown

[symo]

Hi there

 

Things were going swimmingly on my new AMD Ryzen "Threadripper" Win 10 PC - then it started to hang on final/custom renders. It seems to have started doing this after a period of intense rendering - usually on final or custom renders - but occasionally fast renders. Ive ascertained that its not file-specific. Prior to this, the PC would crank through any rendering job I threw at it.

 

Towards the end of the render, after the CPU has been maxed-out on 99% for 10-20 secs, its like the PC just gives up and become completely unresponsive - no mouse movement or anything. The only way out is a hard reboot. The first sign of trouble was a render that ended with all my windows being re-arranged over my 2 screens.

 

After the hangs started happening regularly (within a few hours), I restored the C Drive to an Acronis  backup that was done 24hrs earlier. This didnt fix it.

 

The last time the hang happened, I rebooted to find that my desktop had changed and that the Win 10 start menu wasnt working.

 

My next step is a Win 10 Memory Test - which I have started and will take a couple of hours. There were no problems flagged after 1 hour of starting the test.

 

Can anyone suggest the next steps to trying to resolve this? 

 

Cheers in advance.

 

 

 

Edited July 11, 2019 by symo

[zoomer]

I think that is more a hardware issue.

Bad fit for CPU Cooler, Thermal Paste Problem, bad fit in CPU Socket, Weak Cooler, ....

(All those things happen quite often - but normally from the beginning)

 

I would start with a Monitor App and check CPU Temperatures under load.

If it gets too high I would try demounting the Cooler to check these things and or

reset CPU into it's Socket again from scratch.

 

I don't think it is very likely but Motherboards or RAM can be faulty too.

Older Board generations often have to weak VRM's to run a TR 2990WX, at least

when overclocked. A weak Power Supply can also cause instabilities.

Low Case Airflow could also heat up Mainboard and VRMs too much.

Also I would check for latest Bios and Driver Updates.

[symo]

Many thanks zoomer. Will try a CPU test app next.

[zoomer]

I see you have a 16 core TR (1950 or 2950)

Their power consumption should work fine in any available TR4 Board.

Stock and maybe moderately OC. Don't know if 4,4 GHz is stock or

highly OC.

If you use OC, it could also be the RAM that gets unstable.

[symo]

Hi Zoomer

 

Im not tricky enough to delve into the overclocking world - pretty sure its stock. 

 

I did some CPU load tests and it was fine - got up to around 75 degrees C (I think). Just seems to be VW mid-render that kills it. Really odd - my macbook takes forever to do final renders - but it gets there in the end. Ive boxed the PC back up and sent it to the supplier whilst I go on holiday!

[zoomer]

75°C sounds ok.

 

I see that stock 16 core TRs have 3.4 or 3.5 GHz.

If it has really 4.4 GHz, it is (factory) overclocked.

Maybe a bit too much ....

If you bought a complete machine, I also think it is best

to send it back and ask them about freezing.

 

I have build a Ryzen 2700X machine and haven't seen any problems

with the AMD CPU and VW SLVP calculation or Rendering in RW or C4D

so far. I also heard no complaints from others using VW and TRs.

[symo]

Hi Zoomer

 

Seems you are right:

 

 

Im pretty sure it was running at 4.4GHZ. Will try it running at the base clock speed.

 

Is it a known that a PC will completely freeze then break windows when it reboots - when the CPU becomes unstable?

 

Lastly - I see that the max temp is 68 degrees C. Given that mine reached 75 - does this mean I have a cooling issue??

 

Nice to know there are other with similar setups that are running well.

 

 

[zoomer]

No, everything looks alright.

It is not overclocked.

 

There is always

Base Clock (What you at least should achieve for all cores)

and

Max Boost Clock (When you use only 1 or 2 cores)

and

everything in between. Like in Cinebench, my Clock Speed is

mostly still higher than Base Clock.

 

68°C when rendering multithreaded is also pretty good.

So thermal issues seem unlikely.

You also already ran a Memory Test without any hint of problems.

As you sent it back, the vendor may have more options for

checking hardware problems.

 

10 hours ago, symo said:

Is it a known that a PC will completely freeze then break windows when it reboots - when the CPU becomes unstable?

 

It is always bad when an OS freezes or crashes.

Whatever the reason was.

Could be while trying to write data to SSD and in the worst case

such an unexpected interruption can destroy an operating system.

Which would need a repair or even installing again from scratch.

 

Another idea for VW freezing when RW Rendering would be any

security software like an antivirus that may interfere with VW

and forbid access to important files (?)

[symo]

OK - cheers. All makes sense. Will let you know how I get on.

 

 

[Art V]

On 7/14/2019 at 11:46 AM, zoomer said:

You also already ran a Memory Test without any hint of problems.

As you sent it back, the vendor may have more options for

checking hardware problems.

As @zoomer already mentioned it  is probably a hardware issue.

 

I've lost months of usability due to memory issues and a motherboard defect. First they tested memory which came out ok, then the motherboard got tested which appeared to be faulty and got replaced. The instability/crashing seemed to be solved but after a week it was back. Then they did a longer test run on the memory and found out it was faulty after all so that got replaced. Added additional memory (same brand, type and series) and back were the problems (after a long test run). Al in all it took 5 months to get it all sorted out and then... once everything was running fine when one of those recent infamous patch Tuesday Windows updates bricked the whole thing due to a conflict with security software and I could start reinstalling everything from scratch. Grmbl.

 

Long story short, even if initial test don't show issues it may be a defect that requires more thorough testing to find out. What did solve quite a few pre-hardware trouble issues was that I had to do a complete new install of Windows and that solved some other issues. So you may want to do a refresh of Windows 10, especially if it has had several major updates. You can download the latest install ISO from the Microsoft website to get an as clean as possible install and then install the most recent drivers for your hardware. If it is a software issue after all then this might solve it (no 100% guarantee however).

[symo]

Hi Art V

 

Cheers for that. Ive sent my PC back to the supplier - so I really hope they nail it. 

 

Will reply with any useful info I get.

 

 

[symo]

Got my PC back today and all seems to be well again.

 

Heres what they said they did:

1. Update MB bios
2. Fixed windows folder which was corrupted by the crashes (but not reinstalled win 10)
3. Updated all win updates
4. Wound the CPU clock back from 4.4ghz to 3.5ghz. This was based on an earlier suggestion but they did this anyway. Im loath to crank it back up now its running OK - but may do a bit further down the track. 

Its running great - renders might be slightly slower - but I can live with that. They were adamant that there weren't any hardware issues - but seem to think it might have been a windows update.

 

Ive learnt that some apps give false CPU temp readings - I was getting up to around 76° C mid-render with AI Suite. But have been using Ryzen Master which apparently gives much more accurate readings. Final quality renders are taking it up to 61°C max - so within the 68°C max limit.

 

Adding to my woes were issues trying to restore from a "clean" Acronis backup - which ended-up being corrupted.

 

Now that its running well - Ive disabled win updates as I dont want a repeat experience. Im thinking of doing this periodically - so I have more control of this. Is this recommended?

 

 

 

[Art V]

7 hours ago, symo said:

Now that its running well - Ive disabled win updates as I dont want a repeat experience. Im thinking of doing this periodically - so I have more control of this. Is this recommended?

Short answer: YES!

 

Slightly longer answer, follow this website: https://www.askwoody.com/

 

It will tell you whether it is worth to install the patch Tuesday updates or if there are any major/showstopping issues that should make you want not to install updates for now.

Their general recommendation is to set the updating for patches etc. to 2-3 weeks as most (not all) issues should be resolved within that time frame or updates will be retracted if there are serious issues. That way you'll minimise the risk for thing going wrong. The halfyearly updates should be set to a delay of 365 days if possible.

 

After the patch Tuesday mess I only update if the website says it is ok to do so (i.e. no major issues) though there is always a chance that your setup might contain something that may cause issues anyway with some update. But that applies to any computer system as they can't test all possible combinations of hardware/software.

[symo]

Cheers for that.

 

Might need to do as woody says and upgrade to win 10 pro. 

[Art V]

Win 10 Pro has some advantages over the home version, especially for business use, so I would prefer that over the home version if possible.

[symo]

So after a few days of pain, I finally have a stable VW workstation.

 

Summary:

• It seems that my PC came back from the supplier in a broken state whereby it was missing some basic windows functionality (Microsoft Store). This became apparent when I tried to upgrade to Win Pro and my PC point-blank refused to do so - was missing the store app.
• After a few hours on the phone to MS support (who were actually very good and quick to call me), I reinstalled win 10 a couple of times and then upgraded to pro
• Ive managed to install all the right drivers & software and things seem to be running pretty good - PC performing well in benchmarks. The only thing that didnt work is the Aura app for LED control - but I can live without this.

1. Are there any risks in enabling Windows Restore? I only like to restore from my system from Acronis backups if I really must
2. Are there any other win 10 tweaks I should consider - now that Ive upraded to pro??

Cheers for all the help with this - hoping I get a few good months of pain-free rendering.

 

[Art V]

On 7/26/2019 at 10:41 AM, symo said:

Are there any risks in enabling Windows Restore? I only like to restore from my system from Acronis backups if I really must

No there are no risks in enabling Windows restore, I recommend enabling it. It may be the only way to revert back to a previous state without much damage/work after an update bricks your windows.

 

If you want to rely on Acronis backups then you should create a backup every time you before apply Windows updates or major software installs in case something goes wrong. Otherwise you may lose activations or you may have to reinstall newer software/updates etc. which may take some time as well. This may not be a really feasible option in the longer run. One thing that most likely will fail with Acronis is backing up data that is written to the boot sector by some software you use which may break activation in a bad way (e.g .Autodesk does this or used to to this, and a restore would break the activation of the license and require contacting their support to enable activation again)

 

You may want to take a look at Rollback Rx (https://horizondatasys.com/) as an alternative, it should be able to restore Windows and remove everything installed by an update even if it crashes Windows completely after an update and it may be a more convenient option than Acronis. They do have a free version for home use, but you may want to get the Pro version as it allows for more restore points and some additional options (multiple drives, scheduler etc.).

[symo]

Many thanks for that

 

Edited August 8, 2019 by symo
Didnt mean to quote previous text