-
Notifications
You must be signed in to change notification settings - Fork 151
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
Collect Data on Rule Traces and Performance Across Proof Suite #1048
Comments
Per discussion we went with depth 150 and a timeout of 2000s. If a file is empty then the test likely timed out. Tarball is attached. |
each claim will take about 10-15 K steps, so we should make sure the log for this file gets split into the individual rule traces. Then we want to analyze (in the rest of the proof-suite) how "often" these rule traces occur in the other traces present in the system. |
Here's some output from a first pass at using the new one-line logs to extract information about which steps of a proof are slow. The proof used was |
@JKTKops looks great. It might be interesting to do some different types of passes over the data. Can you try:
|
Also, here is just raw frequency of occurance given the data you posted above:
|
Out of interest in edit: based on Tom's feedback in the meeting today, I think the duration data here is mostly meaningless. However the frequency data would still be meaningful. |
Putting this data through We should look for this trend in the fixed data. |
For reference, this is currently blocked on runtimeverification/haskell-backend#2682 which I'm working on. |
Modify theMakefile
to dump the output of each proof after 100 steps.Inspect the output for unsimplified functions we can add functional lemmas for.See further discussion below.
The text was updated successfully, but these errors were encountered: