Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - LazyViCE

Pages: [1]
1
[Max] Bug Reporting / Re: dual CPU issues?
« on: 2025-02-21, 06:26:46 »
Thank you very much. I am glad to hear that some activity is going on to improve this situation.
I have been using multi-processor assemblies for 4 years or more and I constantly encounter some new problems that I have not encountered before.
Here is another example:
On a system with EPYC 7601 x 2, I encountered the fact that during the benchmark all cores work at 100%, but periodically there is a wild drop. I have seen this on systems with ZEONs, but at least here all cores work in the benchmark.
But when I try to render from 3DS Max on EPYC, only 16 cores work. I had to force the "process group expander" via Process Lasso to work all 128 threads in the render via 3DS Max.
I constantly have to use some kind of crutches to make everything work in an adequate mode.

2
[Max] Bug Reporting / Re: dual CPU issues?
« on: 2025-02-20, 09:47:13 »
All my computers have two processors on board.
Xeon 2696v3 x 2 = 5 System Units
Epyc 7601 x 2 = 1 System Unit
Windows 11 - Enterprise

What problems do I encounter:
* CORONA Benchmark works with drops, the core load drops by 10% every few seconds on all systems
* A similar situation is observed during rendering, but less often. Example:
I am making an animation of a camera flyby in an apartment. In the corridor area, there are constant wild drops and I have no idea what to do with it. I have already tried EVERYTHING! Nothing helps. At the same time, in the hall area, such phenomena are greatly weakened. How to find a relationship, I do not know.
* On Epyc 7601 x 2, rendering from under 3ds max goes ONLY ON 16 CORE! I couldn't solve the problem using Windows, I had to install the Process Lasso program and through it forcibly assign 2 processors in 3ds max. There was no such problem on the Xeon 2696v3.

My Conclusion:
* All software is crooked and cannot fully use the potential of multiprocessor systems
* Corona
* 3ds Max
* Windows
Everything is crooked and needs to be improved, and since all this is an interconnected ecosystem, it will take centuries to fix all this.

3
Thank you for trying to help. I'll also consider the option of replacing Backburner later on.

So! I don't want to stitch the result manually; it's almost impossible to do so. And yes, Backburner should stitch everything automatically, which usually happens.

I seem to be misunderstood... The essence of the problem is clearly outlined in the previous post, along with how to reproduce it. But I'll reiterate:

At the end of the job, I should receive 1 file in CXR format that includes all layers, passes, metadata, and elements.
But this doesn't happen if I add Backburner to the pipeline and enable the "Split Scan Lines" feature.

---
This issue is not new at all. There's already a post where they tried to solve this problem and wrote a script that was supposed to make Backburner save properly, but these workarounds ONLY work with EXR files, and even then, it's far from the best solution.
https://forum.corona-renderer.com/index.php?topic=25686.30

4
I confirm. With the transition to 11 Corona, I started getting random NaN errors. Moreover, they can be caught even on stage in gray material. And the more complex the scene, the longer you work in it, the more often you catch the NaN bug.
The same thing happens in 11.1 Corona. The error is present on all 3 of my computers.

For all those who are trying to catch this bug: I did not find any patterns that would help catch it. It's like random. Relatively speaking, I can work in a complex scene with a lot of suns and artificial light and everything will be ok, but I’ll catch NaN in some kind of closed corridor, which is essentially just a box. And vice versa.

5
Good time of day, everyone!

I spent about 7 hours on the forum trying to figure out and solve my problem. My problem:

1. I have a very heavy scene for which I need 80GB of RAM, and rendering takes a very long time, or 3ds max crashes.
2. The scene contains a large number of render elements, such as LightMIX, and I really need them.
3. I'm trying to render the scene in Backburner with the "Split Scan Lines" feature enabled.
4. As a result, I get several pieces of .cxr files that refuse to properly merge into one .cxr file, as it happens under normal conditions.

To be more specific:

* I get one file that has no passes and is not displayed in terms of lighting as it should.
* and I have, let's say, 5 pieces of images, in which everything is fine, but they do not stitch together. They just lie alongside all the passes and LightMIX, but don't merge into a single .cxr file.
Reproducing the problem is as easy as pie:

1. Create a simple room box.
2. Add some signature teapots.
3. Add light sources.
4. Add light sources to LightMIX.
5. Try to render all this into .cxr as usual, but through Backburner with "Split Scan Lines" enabled.
6. Get a headache.

Software: 3ds max 2024, Backburner 2019 (latest available), Corona 11.1

Is there any way to finally deal with this task?

P.S. There is a script in the some thread that allows saving the .exr format somewhat adequately, but I didn't understand if it's possible to do the same with the .cxr format. With .exr, I barely managed to cope with the script, and even then, it's such a workaround...

Pages: [1]