feat: improve type.hashTreeRoot() using batch #409
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
Performance Report✔️ no performance regression detected Full benchmark results
|
|
tested this on |
|
sha256 works in blocks, each is 64 bytes so perhaps it's more meaningful to reflect that for also with holesky, there are 1.7M validators. For every 8 deposits we have to reallocate the whole 1.7M * 8 bytes = 13.6MB for Update:
|
twoeths
force-pushed
the
te/improve_type_dot_hash_tree_root
branch
2 times, most recently
from
9e32c5c to
7ed3ced
Compare
nazarhussain
force-pushed
the
te/improve_type_dot_hash_tree_root
branch
from
d3821ee to
cbb30a2
Compare
matthewkeil
reviewed
Feb 21, 2025
There was a problem hiding this comment.
I left a few comments but I think this PR really needs to be reviewed by @wemeetagain
| // Merkleization | ||
|
|
||
| protected getRoots(value: ByteArray): Uint8Array[] { | ||
| return splitIntoRootChunks(value); | ||
| protected getBlocksBytes(value: ByteArray): Uint8Array { |
There was a problem hiding this comment.
Im confused by what BlocksBytes represents. I understand there is a shared buffer that is getting reused but why the name Blocks and thus bytes of a block? Perhaps we can talk about the naming conventions here?
There was a problem hiding this comment.
It seems like a Block is the unit used for (2) 32-byte roots?
There was a problem hiding this comment.
Perhaps calling 32 byte chunks Bytes32 and 64-byte "blocks" Bytes64? Not sure, what do you think?
|
|
||
| blocksBuffer.set(value); | ||
| const valueLen = value.length; | ||
| const blockByteLen = Math.ceil(valueLen / 64) * 64; |
There was a problem hiding this comment.
It might be helpful to break out a helper function so its clear why this is happening everywhere.
export function getPaddedByte32Count(buf: ArrayBuffer): number {
return Math.ceil(buf.length / 32);
}
export function getPaddedByte64Count(buf: ArrayBuffer): number {
return Math.ceil(buf.length / 64);
}| const blockDiff = newBlockCount - oldBlockCount; | ||
| const newBlocksBytes = new Uint8Array(blockDiff * 64); | ||
| for (let i = 0; i < blockDiff; i++) { | ||
| this.blockArray.push(newBlocksBytes.subarray(i * 64, (i + 1) * 64)); |
There was a problem hiding this comment.
Why loop and push in chunks? Why not just push all at once and only trigger one resize to the blockArray?
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.
Motivation
Description
getRoots()and compute root from there, this PR implementgetChunkBytes()merkleizeInto()which use batch therechunkBytesBuffermemory in type, almost noUint8Arrayallocations in the middlehashTreeRootInto()api. This is needed in case consumers want to reuse memory allocation thereallocUnsafe()of as-sha256 where it makes sensecherry picked from #378