Fixes #8426: Inline LinearCongruentialGenerator into consistentHash.#8429
Closed
Tsalachsnow wants to merge 1 commit into
Closed
Fixes #8426: Inline LinearCongruentialGenerator into consistentHash.#8429Tsalachsnow wants to merge 1 commit into
Tsalachsnow wants to merge 1 commit into
Conversation
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
consistentHash allocated a new LinearCongruentialGenerator object on every call solely to hold one long and advance it with two arithmetic steps. Under high call frequency this produces a continuous stream of short-lived objects, increasing minor GC pressure for no gain. Replace the allocation with a local long state variable and inline the two arithmetic operations directly into the loop body. The class has no remaining callers and is deleted in full. Output is bit-for-bit identical to the previous implementation: the inlined expressions are the same floating-point operations in the same order. Also add two tests: testConsistentHash_singleBucketAlwaysReturnsZero (buckets=1 must always return 0) and testConsistentHash_largeBucketCountResultInRange (result stays within [0, Integer.MAX_VALUE) for extreme bucket counts).
Tsalachsnow
force-pushed
the
fix/inline-lcg-consistent-hash
branch
from
8d6754a to
9184f66
Compare
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Summary
This PR eliminates the per-call heap allocation in Hashing.consistentHash by inlining the LinearCongruentialGenerator inner class directly into the method body.
Motivation
Hashing.consistentHash(long, int) constructed a new LinearCongruentialGenerator object on every invocation. That class existed solely to hold one long field and perform two arithmetic steps — there was no behavioral complexity to encapsulate. Under high call frequency (e.g. thousands of calls per second in a load balancer hot path) this generates a continuous stream of short-lived objects, increasing minor GC pressure and producing measurable allocation noise in profiler traces.
Inlining the single field and the two arithmetic operations directly into the method as local variables eliminates the allocation entirely with zero behavioral change.
Changes
Hashing.java (JRE): replaces new LinearCongruentialGenerator(input) with a local long state variable; inlines the two arithmetic steps directly into the loop body; deletes the LinearCongruentialGenerator inner class entirely — it has no remaining callers.
HashingTest.java (JRE): adds two new test methods covering previously untested boundary conditions on consistentHash.
Testing
All existing HashingTest cases continue to pass. Two new test methods added:
testConsistentHash_singleBucketAlwaysReturnsZero
testConsistentHash_largeBucketCountResultInRange
Full ./mvnw clean install passes with 0 failures, 0 errors across the entire test suite.
RELNOTES=Hashing: eliminated per-call LinearCongruentialGenerator allocation in consistentHash; output is bit-for-bit identical to the previous implementation.
Fixes #8426