-
Notifications
You must be signed in to change notification settings - Fork 152
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Support for non-immediate proof maintenance #4563
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good. Can you make it toggleable with a flag so we can choose to do the disk writing/status bar updating every step?
Co-authored-by: devops <[email protected]>
…#4566) Adds the number of terminal nodes to `one_line_summary` for `APRProof`. Also changes the display to use `|` as dividers instead of `/`.
Co-authored-by: devops <[email protected]>
Co-authored-by: devops <[email protected]>
This option is unused and all its use cases are subsumed by the `-l Timing --log-format json` non-RPC log option. For reference, here's the Haskell Backend PR that stops emitting the timing RPC log entries: runtimeverification/haskell-backend#4015
I think I prefer to have a separate parameter so that we can decouple it from the number of workers, as it is now. Is there anything else to do in |
You could possibly add the same parameter/logic to |
@@ -1,6 +1,6 @@ | |||
[coverage] | |||
output = "default-file" | |||
definition = "/tmp" | |||
definition = "/var/folders/ks/1p8zyp5j7xz_v1krhl4l17tm0000gn/T" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What is this change for?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have not changed that file manually...
Yes, happy to. |
Just to say - I am not sure how to test this, because the final outcome is the same, it's just that the proof should be faster. Happy to go along with any ideas in a follow-up PR. |
When doing large real-world proofs with a high degree of parallelism, the bandwidth is overtaken by the writing of the proof to disk on every proof iteration, slowing the process considerably.
This PR introduces the ability to write to disk and perform the callback on every
maintenance_iterations
of the parallel proof, set by default to 32.EDIT: Actually, perhaps it's a better idea to write the proof every
max_workers
iterations.