rvgo: experimental no-alloc noescape merkle funcs #88
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Outdated suggestions cannot be applied.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.
Description
A little bit hacky. And makes the assumption there's only 1 concurrent hasher (this could be fixed by allocating a
keccakState
per Memory merkleization job, and passing it as typed function argument through the merkle functions).go:noescape
andgo:linkname
to force it to not assume the function interface, and not allow any slice data to escape to the heapThis essentially removes all the non-essential allocations from the merkleization work, in a bit of an unsafe non-recommended way, to see how much performance can improve with this kind of heap-analysis informed memory optimization.
Profiles of the benchmark show that it's only a little bit faster, even after removing a theorectical ~10 million object allocations.
Before:
After:
CPU:
Before:
After:
Heap-escape:
Thought I'd share this little experiment and the results.