Feat/pull mplex #3
Closed
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
ghost
assigned 
dryajov
force-pushed
the
feat/pull-mplex-3
branch
from
cd93515 to
ea9b678
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Outdated suggestions cannot be applied.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.
UPDATE: This is now ready for review.
TODO:
Bellow are the accompanying PRs:
The current implementation takes around ~15mins more running the mega stress tests defined in interface-stream-muxer.
Here is the outline:
New:
Old:
The issue is here - https://github.com/libp2p/pull-mplex/blob/850bbc52c33038f2cdeff40e83afbcf4823b3895/src/coder.js#L85...L89. I need to rework this part to use a preallocated buffer, instead of allocating it every time.Perf:
I was able to shave off ~15 mins of the new implementations running the mega stress tests, the perf is now about the same (or a little better ;) ) than the stream based implementation.
Archive.zip
The attached zip contains a heap snapshot and a perf log taken with
node --prof --nologfile_per_isolate --logfile=xxxxxx.log --log-timer-events /private/var/folders/_r/6c6jf6m10kb3v9kt45qspw4w0000gn/T/v8profilerProxy.js. The one thing that still irks me a bit is the excessive (compared to the prev implementation) GC time. I think we can improve that a lot by pooling theChannelobjects and reusing them, rather than creating new ones every time.Here is a screenshot from the perf log graph - GC is seems to take ~13% of the total time. (These graphs where generated using the attached logs, and webstorm perf and heap analyzers).