Author Topic: PC crashes when starting IR rendering  (Read 4907 times)

2019-08-12, 09:42:41

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
Hi guys,

I am having some troubles these last weeks, particularly when launching the IR rendering. It makes one-two passes, then the PC goes to black screen without any warning and restarts...

I tried to recover Windows to default and only installed 3ds Max and Corona, but it still happens so I assume this might be something related with the hardware... I have also checked the CPU temperature but everything seems to be fine, never going hotter than 65º on a Threadripper 1950x.

This happened to me in both versions, Corona 3 and 4, and in different scenes, but never happened when rendering without IR.

Note: One of the times the PC crashed I only orbited the model. Could this say anything? Probably the GPU is dying? What would you guys try first?

Thanks!



2019-08-12, 10:31:42
Reply #1

rowmanns

  • Corona Team
  • Active Users
  • ****
  • Posts: 503
    • View Profile
Hi,

Are you running IR with the "Fast preview denoise" selected? If so can you try and disable it and see if the issue still persists?

I'd recommend making sure that all of your device drivers are up to date as well.

Lastly, can you generate a minidump from the crash? This can provide us some information as to where the issue might be. Instructions on how to do this are in my signiture.

Thanks,

Rowan
Please read this before reporting bugs:
How to report issues to us!

2019-08-12, 10:35:55
Reply #2

Juraj Talcik

  • Moderator
  • Active Users
  • ***
  • Posts: 3699
  • Tinkering away
    • View Profile
    • studio website
3dsMax crashes are Corona related, but full PC crashes are almost always hardware related so this will probably have little to do with Corona.

But let me give you few tips:

- You ask correct question about GPU, but it doesn't mean it's dying. Corona IR right now by default starts with Optix denoising run by GPU. So when you start IR, both your CPU and GPU will at some point work almost 100perc, putting lot of stress on your power cascade (VRM,etc.), cooling and PSU. CPU temperatures can be ok, but you have to check the temperatures of everything else as well.
- But then you write this happens with simple orbiting. Viewport only uses GPU (easily up to 100perc at moments) and single-core. This alone shouldn't stress either power cascade or PSU.

Nonetheless, what is your motheboard ? what is your PSU ? (Write Case and CPU cooling too).

Here bit of trouble shooting routine, it sucks but usually you have to go step by step:

- Install latest board UEFI and chipset drivers. Then install newer stable (non-beta) GPU driver available.
- Run memtest. Lot of crashes are memory related. If anything comes by, try reseating them first. AMD UEFI updates include microcode (AGESA) that improves compatibility and stability of memory.
- Run latest Prime95, it has AVX instruction set and will simulate Corona...decently enough. It's one of the best stability tests. Your PC should survive overnight at least :- ).
- If you suspect GPU the most, try to replace it temporarily. Borrow from friend, or use some 50 euro cheap one. Don't forget to turn off Optix :- ).
talcikdemovicova.com  Website and blog
be.net/jurajtalcik   Our studio Behance portfolio
Instagram   Our studio Instagram, managed by Veronika

2019-08-12, 12:35:27
Reply #3

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
Guys, thanks for your help!

I followed the steps suggested by Juraj:

I don't really know what did the trick but I updated GPU, UEFI and chipset drivers, and everything seems to be working now! (fingers crossed...!)

Will keep you posted if this happens again.

Thanks!

2019-09-10, 08:24:16
Reply #4

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
Hi guys,

It happened again...!

Tried everything Juraj mentioned, all working as it should, and then I run Furmark to check the GPU health. PC restarts 1-2 seconds after starting the test, so I guess the GPU is the culprit here.

Would you try any other thing before ordering a new one? Also, which GPU would you recommend me under €400? Anything to be released that is worth waiting?

Thanks again :)

2019-09-10, 17:19:39
Reply #5

Juraj Talcik

  • Moderator
  • Active Users
  • ***
  • Posts: 3699
  • Tinkering away
    • View Profile
    • studio website
What is your GPU ? 1-2 seconds after starting the test is odd.. but dead GPUs is nothing unusual even in this age. These things run ultra-hot.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik   Our studio Behance portfolio
Instagram   Our studio Instagram, managed by Veronika

2019-09-10, 17:51:40
Reply #6

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
It's an old gtx 970. What is more confusing is that I'm now testing it on my old computer and seems to be working fine... even it runs Furmark without any problem. Any ideas?

2019-09-11, 07:40:21
Reply #7

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
Update: When I run Furmark, the screen always go black and I realized that the VGA light turns on in the motherboard. Does this give any clue?

2019-09-20, 19:51:55
Reply #8

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
To the ones that wanna know what this was all about... the graphic card was dead so I ordered a RTX 2060 Super from MSI. I will test it on Monday and will let you know.

Thank you all!

2019-09-27, 11:49:19
Reply #9

letsbreaktherules

  • Active Users
  • **
  • Posts: 9
    • View Profile
Update: The problem persisted after installing the new GPU. Then, I downgraded the BIOS to the first one that supported Threadrippers (yes, I've said downgrade), and everything is working fine so far. Will let you know if this changes.