search

slic3r / Slic3r

Open Source toolpath generator for 3D printers
https://slic3r.org/
GNU Affero General Public License v3.0
3.33k stars 1.29k forks source link

slic3r hangs forever with 200% CPU when attempting to slice #5019

Closed jin-eld closed 3 years ago

jin-eld commented 4 years ago

I have a model that does not slice properly if particular settings were applied. I did not identify what settings exactly influence it, but I am attaching a working and a non working .ini

This happens on master HEAD (92abbc42dfdd5385c1f9c3a450e2f3da835f8b8d)

Describe the bug

Load the model and slice it - the progress dialogue will appear and just hang there, CPU load goes up to maximum. I tried waiting for a while, but slic3r did not recover.

This does not happen on slic3r-1.3.0-13.fc31.x86_64 - there the model with the given settings does slice correctly.

3D Model and Slic3r Configuration Export bug.zip

Expected behavior Successful slicing to a g-code file.

Desktop (please complete the following information):

Additional context

working.ini does slice the model on the master HEAD revision non-working.ini exposes the error

non-working.ini slices fine with the released 1.3.0 version

dwillmore commented 4 years ago

Wow, I'm looking through the differences and I'm just not seeing anything that sticks out. Anyone know a quick way to do a binary merge?

jin-eld commented 4 years ago

I think I know what causes it. When I have "Minimum shell thickness" set to zero, slicing works. When I set both values (vertical and horizontal) to 2mm - it will hang when slicing.

dwillmore commented 4 years ago

Verified here. Used Slic3r-1.3.1-dev-92abbc4-x86_64.AppImage on Fedora 32 with my profile (similar to @jin-eld's profile) and it hangs. Restarted it, changed both shell thicknesses to 0 and resliced. Worked just fine. Changed them back and resliced--hang.

It's just a bisect to find out the commit. I wonder if the cli version has this problem as well. I've never used it, but it would be a lot easier to use it to bisect than the gui.

joluxer commented 3 years ago

Found a simelar bug here, where the slicer hangs, when the horizontal shell thickness is not an integral multiple of the layer height and sometimes also also when the vertical shell thickness is not an integral multiple of the extrusion width.

lordofhyphens commented 3 years ago

I found a bug in the horizontal (and you can trigger it with the CLI using --load).

The feature was computing std::abs(min_shell_thickness - current_shell_thickness) > epsilon as a loop conditional, which meant that if you had an oddball layer height it could shoot right over -epsilon < 0 < epsilon and then continue forever.

lordofhyphens commented 3 years ago

@jin-eld @dwillmore There was an issue around the top/bottom minimum shell thickness where the check for shell thickness can skip completely over the threshold (epsilon).

There's a fix now available in a PR.