V 19.1 & 18.6.6 - System gets stuck after undefined renderi

After years I have a new power workstation, in principle similar to the old one, of course the components are now even more powerful.
I first installed Davinci V 19.1 Studio (beta), but got the following error.
I am now back to the latest version 18.8.6. Unfortunately the same thing is happening.
Looks like it's the interaction between the system and Resolve, but I have no idea what to do (I'm not a "computer specialist").

When rendering projects (1920x1080, default Vimeo settings here) that have already been rendered dozens of times without any problems (with the old workstation), the system stops completely after some undefined time.
(e.g. film of approx. 90 min length: after 80% rendering time, at 98% or at 45% - a shorter similar video of 20 min was rendered without any problems).
A cold start is needed.
This phenomenon only occurs in conjunction with Resolve.
It looks as if it is due to the interaction between the system and Resolve, because otherwise the workstation runs with all other programmes all day without failures.
But I have no idea what to do (I'm not a "computer specialist").

There is a Resolve crash report available for viewing:
https://icedrive.net/s/QbGxjZfSXVDxFxgxyNXXYz42hvzA

Maybe someone can help or has an idea?
Thanks!

System:
Intel(R) Core(TM) i9-7940X CPU @ 3.10GHz, 3096 MHz, 14 core(s), 28 logical processor(s)
(RAM) 32.0 GB
Windows 10 Pro, Version 10.0.19045 Build 19045 (current)
NVidia: Geforce RTX 4060 (driver: V 552.22 Studio - current)

Sounds like an overheating issue that goes away when you restart. When is the last time you repasted that CPU? Its quite an old gen. Rendering in small chunks might alleviate the problem to an extent and then longer renders with the CPU being pegged all the time raises temperature.

Or faulty RAM.

Hi,

Its a new work station < 4 weeks old ... so no CPU pasting thing ... and no heating problem ... temperature inside is low ... fans do not to start to rise speed before crash ...

I have to find out the real cause ... but how? What does the crash report says?

Not sure what the logs say. I'm just making observations based on your info. Computer freezes and reboots are not usually software related...

...e.g. film of approx. 90 min length: after 80% rendering time, at 98% or at 45% - a shorter similar video of 20 min was rendered without any problems).

I think Jim is on to a solution to at least let you get the project done:

• render the film in 20-minute segments to a near-lossless mezzanine format (DNxHR 444)

• let the system cool down and/or reboot inbetween segments

• once you're done create a new (empty) timeline with no corrections or operations

• join all the clips together in one continuous timeline and render it out to the format of your choice.

This was standard procedure for us for five years with the (eternally damned) 2013 MacPro, which basically had overheating GPUs. Once we got the 2019 MacPro and the later Mac Studios, this problem went away. (We do tend to render features at no higher than 25fps render speed.)

We never use the "default" Vimeo upload specs, because Vimeo / YouTube / Facebook / Instagram all re-encode their videos anyway. We just use the correct size and framerate and upload in a benign format like ProRes 422LT. I think DNxHR HQ would be close.

Do that and see how it looks. Actually isolating down the source of the slowdowns and crashes would take hardware, OS, and driver troubleshooting. Generating a Diagnostics Log and getting it to BMD Support would be my first suggestion there.

Hi Marc,
thanks for the answer.
Maybe I have to go that way for the time being.
The workstation manufactorer is looking about hardware and/or system problems now. Will see!

Anyhow there is absolutly no heat problem at all. The problem even starts within 8 to 15 min. rendering on a "cold" machine. Additionally this is a workstation for "heavy load" purpose (CAD etc.) and has inside big cooling towers, several fans and control functions.

I think Jim is on to a solution to at least let you get the project done:

What's a Jim? ).

Anyhow there is absolutly no heat problem at all. The problem even starts within 8 to 15 min. rendering on a "cold" machine. Additionally this is a workstation for "heavy load" purpose (CAD etc.) and has inside big cooling towers, several fans and control functions.

How are you making this assessment? Have you monitored your system with something like HWINFO64 and check the temperatures there?

The case having big tower coolers and etc doesn't mean the CPU doesn't get hot. Especially if its not been pasted properly and the tower cooler doesn't make surface contact properly.

Or if the fans are not installed properly and they just don't get the heat out of the case.

Good luck.

Yes, tested with CoreTemp ... max. 71°C ... no heat problem ...

You've probably seen this recent post already but if not, perhaps there are similarities:
viewtopic.php?f=21&t=198527

I've seen this on my system for a long time but I have to say it's got better with the last few versions.

It usually never happens following a reboot. But if I work with DR for hours and then attempt a render, there is a good chance the render might hung OR will be very slow.

HOWEVER, the whole OS is still usable, it's only Resolve. So if your whole OS freezes, there might be an issue here.

Because you have DR Studio, check the GPU temperature - the "hotspot" in particular. Studio uses the GPU more than the CPU and I wouldn't be too surprised that a new GPU might have overheating problems. It might not be, but takes a moment to check - but you need to use HWinfo or similar, not just by listening the fans or looking at the cooling towers! You said you're not an expert on HW, trust us then

If all checks out, a fresh Windows install might help? I know it's not ideal though.

How are you making this assessment? Have you monitored your system with something like HWINFO64 and check the temperatures there?

The case having big tower coolers and etc doesn't mean the CPU doesn't get hot. Especially if its not been pasted properly and the tower cooler doesn't make surface contact properly.

Jim, typically CPU's and GPU's throttle down if overheating.

HR Lehner, did you do a new windows install when you moved to new system or you put old drive in new computer, because that might cause some problems with drivers especially Nvidia GPU drivers. Would want to use DDU to get rid of any old GPU drivers and reinstall Nvidia drivers.

Resolve can crash the drivers and make system unstable while other software won't cause a problem. I've seen it happen on timeline, and it's unrelated to heavy GPU use, infact I had the idea it could be more related to the GPU hitting higher clock frequencies because it wasn't under full load which causes the problem. Not exactly the same problem as you, but something else to look at especially if you did not do a new windows install for new computer.

Hi

From your logs you are running an HEDT platform based around an Intel Core i9-7940X.

I also run an HEDT platform based around the 10980XE and had a similar issue back when I moved to the platform.

I traced the issued back to an 'Auto / Default' BIOS setting and AVX. Most motherboard manufacturers on overclocking CPU SKUs, which yours is, will technically run the CPU outside of Intel Specs by having the Single Core Boost Clock run on All Cores including AVX.

ASUS for example calls this 'ASUS Multicore Enhancement'.

I found under demanding workloads the motherboard was not delivering enough voltage to the CPU. I manually set in the BIOS the All Core Clock to the Single Core Boost clock but set an Offset for AVX so that when AVX was used it ran All Core Clocks at Base Clock speed.

I also tweaked the Voltage via an Offset to reduce thermal load.

I would try running the system with the BISO set to run the CPU at Stock Values and see if you have the issue. If not you can then tweak the BIOS settings to get better performance until it crashes again.

I spent the better part of day tweaking my CPU so I was able to tweak the best performance for the lowest heat. Only downside is the CPU now pulls 350W under load.

Regards

Peter

Thanks to all your answers.
- Its a complete new system incl. a new version of Win 10
- the hardware manufacterer got the results of the maschine now. Some more tests are necessary - if the results do not lead to a solution they will replace the graphic card or the whole machine, will see next week ...
- BIOS modifications maybe could be the next thing to try, if ...

Jim, typically CPU's and GPU's throttle down if overheating.

I keep asking what or who Jim is and since you quoted me I'd say get your lenses checked since my name is Andrew.

And secondly, no. If improperly pasted CPUs and GPUs will artefact and then freeze and crash. Same thing if the fans inside a case dont get proper airflow and keep recirculating hot air. Once a temp will go above a certain level (above TJMax) for a time the cpus have sensors that will shut down to prevent damage.

And this is on a properly pasted but inadequately cooled system. The improperly pasted ones will have hotspots that could trigger freezes even if the monitoring software says the temperature is fine.

And running too much voltage can also do that as Peter said above. And you can also run too much voltage through the SoC (memory controller, IO) and also get shutdowns and freezes regardless of temperature.

If throttling was the only outcome out of any of this, what a wonderful world it would be.