Machine cold boots after enabling shadows.

I cannot play with shadows enabled, anywhere from 1 - 10 minutes later the pc restarts.
You can even just leave it up on screen and don't even manipulate the environment / move the character
and it will crash without fail.

I cannot really /bug this unless logging in after the reboot and /bug'ing it will help at all?

I'm now currently using a Gigabyte x399 sli plus motherboard, Threadripper 1950x and 32 gb of ram. I use a MSI Nvidia GTX 1070ti with the drivers version being 397.55 which is the version that had the hotfix with the gpu hitting 99%.

This rebooting with shadows has been happening since release 53. It started happening directly after a release not a driver upgrade.

There are no errors in the event logs, especially regarding unity, sota or nvidia.

I have tried stripping all nvidia drivers, installing latest.
I have repaired my installation of SOTA.
I have removed it entirely and deleted all the game files associated and reinstalled it as well.

 

Didn’t you already post this?

 

Instant reboot = PC overheating. Check your temps when doing that to be sure this is the case.

 

I've stress tested with 3dmark 4K renders. no reboots.

The machine is liquid cooled at 30C ish.

 

Can you check your blue screen options. As if you tell it not to do a dump you can miss them

 

Liquid cooled CPU I presume. Try disabling as liz_the_wiz said. Also a nice tool to use hwinfo: https://www.hwinfo.com/ Have that open while you enable shadows and see what is exactly happening.

 

I will try that app. There is no blue screen, just boop restart.

I'll see if the app shows anything interesting.

 

Here's what i got ... next step is to let it crash. As far as i could tell the stats were stable / not increasing. In fact the cpu temps dropped like 3 degrees with shadows.

 

It seems odd that windows is not blue screening but restarting.. that would strike more as power supply - can you please check eventlog for any system messages just before restart

 

As stated there are zero errors or warnings during that time frame.
Not even an unexpected shutdown.

This is the only game that does this. 3DMark 1k/2k/4k stress tests all complete without error or resetting.

I'm pretty confused.

Power supply is a EVGA SuperNova 850P2 850watt psu.

I have one m.2 drive, 32gb of ram (pulls some power) and the threadripper as well as a 1070ti that basically idles in most games.
Is there an app to monitor the PSU usage / rail limits?

 

This was my first thought, the power supply. Some are more prone to fail (slowly) over time more than others and it isn't always consistent in the beginning. I also wonder if it is BIOS related but that would be a bit beyond my realm of expertise on machines I'm not familiar with

 

I'm hoping an EVGA platinum power supply is not failing after 5 months of service. Keep in mind, much higher draws were put through during
the 1k/2k/4k stress tests. Nothing else has crashed or reset this computer. However I'm willing to give it a good test if anyone else has a better
way of testing it.

 

Yeah the temps seem perfectly fine/safe. By the way, drool. You have an impressive rig

 

For giggles I updated the BIOS, sound drivers, and chipset driver for my motherboard. I also put int Nvidias new Driver 397.64.
Still doin the reboot thing.

 

Thanks I suffered with a tablet for like 4 years haha. I knew this would a 5 year + computer so I tried to build it nice.

 

Fun fact, maximum Shadows enabled all day long in Ubuntu Linux 18.04 nvidia drivers 390.59 on both the Live R53 client and the QA54 client, no reboots.

 

NO! It isnt a overheating case. My temp on the CPU page is from 30-60C when the freeze and reboot accurs. The GPU is around 50C.
I don't know what OS this thread maker uses, but this is a ballpark of GPU-locks. Not a throttle problem.

 

Interesting

I have the same rebooting problems with this:
8x Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz
32862MB (32042MB used)
Linux Mint 18.3 Sylvia
GeForce GTX 1050 Ti/PCIe/SSE2
X11 Vendor The X.Org Foundation
4.5.0 NVIDIA 384.111

I wonder if it is viable to upgrade to 390.59 drivers even if it isn't supported by my kernel.

 

I grabbed em from Nvidia's site. It they don't work for you, just apt purge em. You should be able to upgrade them via apt even to 390 ish.

What kernel are you running?

 

Also are you using m.2 chips or SSD drive es? Or both?