Optimize occurrence analysis #1328
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…y computed, instead of re-traversing the term. Pull out a little common machinery between this and DCE, where we also do this optimization. On the BFGS example, this speeds up occurrence analysis by 60-70%, and (according to the profiler) is worth something like a 7% improvement to compilation time end-to-end. Also make two minor improvements to classes relating to WriterT1 that came up while experimenting with this.
`censored` is inspired by `censor` from `MonadWriter`, but (i) can't reuse that name, and (ii) adverbs are better here anyway because we are modifying actions.
axch
force-pushed
the
optimize-occurrence-analysis
branch
from
ff3393d
to
aa8a786
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
It was quadratic, because it would recompute the free variables of the rest of the block at every decl. Now it doesn't do that, using the same free variables trick that DCE uses for the same purpose.
On the BFGS example, this speeds up occurrence analysis itself by 60-70%, and (according to the profiler) is worth something like a 7% improvement to compilation time end-to-end.
The implementation was a little subtle because I had to make sure that the
Accessmap's keys were exactly the free variables, without extras. Before this, extras were ok, because names were just looked up in there, but now the absence of a name is also meaningful, so has to be correct.