Open pupeno opened 2 months ago
I think the the problem is triggered by the shelled instead of hollow option.
I'll try to reproduce. Just wondering, are you doing that with base magnet sockets/screw holes enabled or disabled? Asking because that could be the source of square-ish complexity for pattern feature, screw holes are created as pattern of 4 elements and then base tiles would be generated as another pattern feature including the previous one inside it. I hope it's something else though, but can't think of anything else right now
No magnets or screw holes:
This is what it looks like when it gets frozen:
tried to reproduce, I can see that "shelled" option takes quite a bit longer comparing to "hollow", but it still succeeds for me within a few seconds. When I increase the size up to 10 by 10 it takes about 6-7 seconds for "hollow" and ~13 seconds for shelled version which is almost 2 times difference (in the timeline I can see that it spends a lot of time on shell operation specifically). I wonder if using "hollow" option would be able compute in your environment or if it would also freeze
Hollow works, it's only shelled that gets stuck, and I have let it running for a few minutes. The CPU I'm running on is a Intel Core i7-6850K 3.6 GHz, and the graphics card is a nVidia GeForce RTX 3090.
I was trying it on an m3, will give it a try on a windows machine when I get a chance. To my knowledge fusion neither takes advantage of multi core nor does it use gpu for model computations.
I suspect the only way to resolve this is to implement shell feature manually by building and subtracting inner volumes
tried it on Win 11, with Ryzen 5 3600x (I think it's 4.4 GHz), it didn't freeze while generating this 10x10 shelled bin, took a bit longer, just under 25 seconds.
I'm running on Windows 10, I doubt that makes a difference though.
Is there anything I can do to help debug this? And if not, if it's not reproduceable, I think just closing it is ok.
With the bin length of 3 it worked well, with 5 it took a while but it generated the bin, maybe a minute. With bin length 8 it's just frozen. If it's ever going to finish, the time complexity feels exponential or higher. Could it be something triggered by making the bin size 12mm by 12mm? Or shelled?