kimimaro
kimimaro copied to clipboard
seung-lab
Progress bar stopping with parallel
Hello, I have multiple cores, but when choosing parallel = 4 or parallel = 4 to compute the skeletons, the progress bar stopped at some point. It only goes to the end when using parallel = 1.
Hi Alexandre,
If you can do me a favor, I've pushed a branch wms_spawn_mp which switches the default process spawning method from fork to "spawn". Give that a try and let me know if that fixes things. Sometimes "fork" can deadlock on shared resources.
Can you also let me know your OS and Python version?
Will
Hi, Thank you for your answer and you help. It seems to have worked on some trials, but now running on the full dataset it still stops at some points.
I am on Ubuntu and python 3.8
Update:
I did work on some trials. Sometimes it still stops, but sometimes it does go until the end.
Hi Alexandre,
I'm currently on vacation so I can't investigate this further right now, but this issue has bedeviled me for some time across many programs. My best guess was the fork vs spawn issue. I'm not really sure where to go from there. Maybe you can find something interesting in the dmesg log during a freeze?
On Thu, Feb 2, 2023 at 12:50 AM Alexandre Di Piazza < @.***> wrote:
Update:
I did work on some trials. Sometimes it still stops, but sometimes it does go until the end.
— Reply to this email directly, view it on GitHub https://github.com/seung-lab/kimimaro/issues/78#issuecomment-1412591048, or unsubscribe https://github.com/notifications/unsubscribe-auth/AATGQSNHXR7GJZZT2U7NML3WVKZPJANCNFSM6AAAAAAUKMPYGM . You are receiving this because you commented.Message ID: @.***>
Hi,
Don't worry, thank you for your help. It worked some times and it was enough to provide me the results I wanted. Thanks for your work!
Sorry I can't be of more assistance right now. I'm glad it worked out. One day I will figure out this bug...
On Thu, Feb 2, 2023 at 7:41 PM Alexandre Di Piazza @.***> wrote:
Hi,
Don't worry, thank you for your help. It worked some times and it was enough to provide me the results I wanted. Thanks for your work!
— Reply to this email directly, view it on GitHub https://github.com/seung-lab/kimimaro/issues/78#issuecomment-1413808331, or unsubscribe https://github.com/notifications/unsubscribe-auth/AATGQSPYTVBWSRJX7UVHVKDWVO6CZANCNFSM6AAAAAAUKMPYGM . You are receiving this because you commented.Message ID: @.***>