You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
right now we are setting queue sizes in the range of: action_queue = 64, and diffuse_queue= 2048 (or 4096). It works fine until around RMAT-22 size graphs but then as large graphs are run the queues get full and at some point they deadlock. One solution is to have large queue sizes but that is really not a good solution, how large can we go? So perhaps find a graceful way of dealing this deadlock.
The text was updated successfully, but these errors were encountered:
I have implemented pruning that helps with deleting actions that are not needed. Also we can increate the queue sizes a bit to overcome the issue. Right now that suffices but will keep this issue open for future thinking.
Investigate how we solve deadlock on queues?
right now we are setting queue sizes in the range of:
action_queue = 64
, anddiffuse_queue= 2048 (or 4096)
. It works fine until around RMAT-22 size graphs but then as large graphs are run the queues get full and at some point they deadlock. One solution is to have large queue sizes but that is really not a good solution, how large can we go? So perhaps find a graceful way of dealing this deadlock.The text was updated successfully, but these errors were encountered: