Computer totaly Stop while using animated procedural cloud

[johan_hugele]

Hello Team,
I have a (big) problem, i often had some strange stop of my computer, something like my both screen turn blue, then black, like if my computer sudently die ...!
and cant reboot exept by doing a hard reboot by turning off then on my power supply button and start again.
I am a Vray and vantage user, this is my everyday work, and i finally found that this issue ( that happen not so often) happened only while using vray IPR rendering or when using chaos vantage, and today i was playing with animated cloud in vantage, and this issue happened to me again.
I hard rebooted my computer, opened vantage again, playing with animated cloud and computer crashed again very quickly.
It is strange as i just finished a production moovie render with vantage, forest and anima etc, and i had not any crash.
it seem thats only the animated clouds make it crash.
Do you have any idea of what it could be ?
I am using the last vray version and the last vantage too, running on a RTX 4090 and a 2990WX thread ripper.
i am just wondering if this issue is about my hardware or maybe something like nvdia driver version or if it come from a problem with procedural cloud.
Hope you can help me about that.
thanks

[johan_hugele]

hi again, well i read that i am not the only one having this black screen problem, seem that the problem come from my 2990WX processor and mother board and ram configuration etc...
ive read on chaos corona forum that other have the same issue when rendering.. i ll probably need to change my computer specs, unfortunatly...!
I made many test to try to understand what is the problem and the issue happen every time i use this procedural animated cloud in vantage, but everything else working good, i can render heavy scene for hours in vantage without any black screen.
Question is , is there anything particular with those animated clouds about RAM usage or CPU/GPU usage that could make this issue happening everytime im using it ?

[npg]

cant reboot exept by doing a hard reboot by turning off then on my power supply button and start again

I had an issue with this same symptom some time ago, only when loading a specific scene in Vantage and it didn't reboot every time. The issue was the power supply. GPUs (especially recent ones) can cause big spikes in power usage for a small part of a second, enough to trip the over-current or over-power protection of your PSU. This is why the reset button doesn't work and it requires a full power off to boot again. If you're curious for more details watch this https://www.youtube.com/watch?v=wnRyyCsuHFQ

I doubt the 2990WX is to blame in this scenario because Vantage doesn't use a lot of CPU except during initial scene loading. (I'm not saying it can't be causing other issues though).

For such a powerful machine you probably need over 1000W of power supply to ensure stability. What is your current PSU?

[johan_hugele]

Hi and thanks for answer, ill have a look at this video link !
Actually my PSU is a 1000W modular one.
Do you think it is not enought and could be the source of my problem ?

[npg]

Yes, I think this is the most likely reason unfortunately. Note that TDP is averaged over time, so the 4090 may indeed be averaging up to 450W, but the transient spikes could be up to twice this power. Even though 1000W should suffice on paper, maybe the combination with the Threadripper and/or not the best VRM on the 4090 PCB... I'm speculating

Of course, you shouldn't just go off and buy a new PSU on my word only. I think one of the easiest ways to verify would be to install MSI Afterburner and use it to set a lower power limit and/or clocks for the GPU. It's probably right around the limits of the PSU if it crashes only under special conditions, so I hope this would prove the theory.

[johan_hugele]

So i installed MSI afterburner and reduced the power limit to 80% then 60 %, i also tried to lower the Clock and then tried Vantage again, but unfortunalty i get a crash everytime again.
I did the test almost 10 times with differents MSI AB settings but nothing change.
In fact im not at all an overclocking specialist so i maybe missed something but i think i made the right things.
Something is very strange that the problem happen only in one case :

I open vantage, enable a sunlight, set the environnement mode as physical sky, enable clouds and ground shadows, set sun position mode as animated geolocation.

Then if i moove the x or y offset, i can play with the cloud position, i can moove slider fast, viewport react fast and clouds moove fast as well and no crash

But if i use the " set start keyframe " then change hour and "set end keyframes " and use the play button to make the clouds moove , i get a crash everytime and after like 1 or 3 seconde maximum. ( i mean this black screen and non responsive computer )

Everything else work perfecty, that makes me crazy ...!

i just dont understand why this specific play button with animated cloud would make my PSU or GPU crash everytime while everything else work fine ?

If you have any other idea of what i could try ?
And if you are right about my 1000w PSU, then i will buy a newer one.

[npg]

We will try to reproduce the issue with the described steps.

I think there are two mutually exclusive possible issues: 1) power issue; 2) bug in Vantage that causes the device to hang/crash. The thing is, if it was #2, then it shouldn't be a problem to reset the computer with the reset button instead of having to fully cycle the power.

Silly question, but did you click the apply button in MSI Afterburner? It's like a checkbox at the bottom.

[johan_hugele]

answer to the silly question is yes ( but i can unserstand you ask it in case haha ), i applied the change with the button in MSI AB, i also created a user profile to be sure.

the issue happen when i do same as the offical video tutorial https://www.youtube.com/watch?v=reRF2LlY6jY at the 2.50 time .
It crash when clicking the play button, or playing the timeline manually.
I would love to know if it is Home or Forum problem, then i can maybe think or not to change my PSU.
Anyway thanks for your time

[johan_hugele]

Sorry not home* or forum*

I would love to know if it is 1) or 2) problem, then i can maybe think or not to change my PSU.

[johan_hugele]

can confirm that my reset button dont work after the crash, but the reset button work normaly as i just tested it.
So i guess i should start to look what new PSU i should buy..!
Do you think i should go for 1300W, 1500 W ?

[npg]

I tried the steps that you described. It didn't crash, but I noticed something I had forgotten about. As I said Vantage uses very little CPU once the scene is loaded. But if you change a parameter that affects the sky model, such as the time of day, we recompute the sky *on the CPU*. This is why it is a bit slow when you do that (especially with the Prague (PRG) model). This is another clue that points to a power supply issue because your Threadripper must be using a lot of power when you animate the sky and the GPU is also actively rendering the scene at the same time.

On the other hand, you said you read that other people have similar problems with similar Threadrippers/motherboards/etc. So I would say that with the new information we still have two possibilities, but # 2 becomes the CPU/motherboard.

Similar to what you tried with Afterburner, you could limit the threads that Vantage uses from the task manager, go to Details, right click on Vantage and "Set Affinity". Select a small number of threads like 2 to 4 and see what happens. You can watch the Performance tab to verify the CPU load (note that GPU load is not very reliable, it's better to use GPU-Z for that)

[johan_hugele]

I tried to disable some cpu cores, tried with 1, 2, 5, 6 but still crash..
i get 2 option, buy a new computer or stop using those animated cloud that my computer dont like much ...!
i will also try to unplug some RAM as i get 8 stick of memory for a total of 96 GB, maybe it can come from this..

computer makes me crazy sometimes ..
anyway thanks for your help =)

[johan_hugele]

last question, im running the nvidia driver 531.41, i just seen that there is the 532.03, do you think it could make the différence ?

[npg]

No, I don't think this is related to the driver. But it's not a bad idea to keep it up to date. The latest game-ready driver is actually 536.23 and latest studio is 535.98.

[johan_hugele]

Hello again,
i tryied to understand again whats going on with my computer and still didnt changed my PSU.
anyway, by doing the animated cloud test again and again, i finally found that the issue happened only with the PRG clear sky model, (the one by default), computer crash always in about 2/3 second when playing the time line.
But i finally tried one other sky model, Hosek et al, and every other works fine and dont crash, i can play the animated cloud fine and no crash anymore.
and trust me i did the test more than 30 times.
Question is, when you tried to reproduce, did you used the PRG clear sky model as well ?

i just wonder about what you tell me with my PSU, or maybe my hardware settings, but it is strange that my computer crash only (and everytime) while doing this particular cloud scenario right ?
Is the PRG clear sky is asking more GPU or CPU than the other ? this PRG would make my PSU overload more that Hosek one ?
Sorry for all my asking, maybe i should just not use this PRG sky anyway but i try to understand what is going on.
Thanks by advance.