Skip to content

perf: improve varint writing performance #618

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
anthony-swirldslabs merged 1 commit into from
Sep 18, 2025
Merged

perf: improve varint writing performance #618

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 5 additions & 27 deletions pbj-core/pbj-runtime/src/main/java/com/hedera/pbj/runtime/ProtoWriterTools.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1010,12 +1010,7 @@ public static int sizeOfVarInt32(final int value) {
* @return the number of bytes for encoded value
*/
public static int sizeOfVarInt64(final long value) {
if (value >= 0) {
return sizeOfUnsignedVarInt64(value);
} else {
// Must sign-extend.
return MAX_VARINT_SIZE;
}
return sizeOfUnsignedVarInt64(value);
}

/**
Expand All @@ -1025,11 +1020,7 @@ public static int sizeOfVarInt64(final long value) {
* @return the number of bytes for encoded value
*/
public static int sizeOfUnsignedVarInt32(final int value) {
if ((value & (~0 << 7)) == 0) return 1;
if ((value & (~0 << 14)) == 0) return 2;
if ((value & (~0 << 21)) == 0) return 3;
if ((value & (~0 << 28)) == 0) return 4;
return 5;
return sizeOfUnsignedVarInt64(value);
}

/**
Expand All @@ -1039,23 +1030,10 @@ public static int sizeOfUnsignedVarInt32(final int value) {
* @return the number of bytes for encoded value
*/
static int sizeOfUnsignedVarInt64(long value) {
// handle two popular special cases up front ...
// handle popular special case up front
if ((value & (~0L << 7)) == 0L) return 1;
if (value < 0L) return 10;
// ... leaving us with 8 remaining, which we can divide and conquer
int n = 2;
if ((value & (~0L << 35)) != 0L) {
n += 4;
value >>>= 28;
}
if ((value & (~0L << 21)) != 0L) {
n += 2;
value >>>= 14;
}
if ((value & (~0L << 14)) != 0L) {
n += 1;
}
return n;
final int clz = Long.numberOfLeadingZeros(value);
return ((Long.SIZE * 9 + (1 << 6)) - (clz * 9)) >>> 6;
}

/**
Expand Down
95 changes: 95 additions & 0 deletions ...gration-tests/src/jmh/java/com/hedera/pbj/integration/jmh/varint/VarInt32SizeOfBench.java
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,95 @@
// SPDX-License-Identifier: Apache-2.0
package com.hedera.pbj.integration.jmh.varint;

import com.hedera.pbj.runtime.ProtoWriterTools;
import java.io.IOException;
import java.util.Random;
import java.util.concurrent.TimeUnit;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.CompilerControl;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OperationsPerInvocation;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;
import org.openjdk.jmh.infra.Blackhole;
import org.openjdk.jmh.runner.Runner;
import org.openjdk.jmh.runner.options.Options;
import org.openjdk.jmh.runner.options.OptionsBuilder;

@SuppressWarnings("unused")
@State(Scope.Benchmark)
@Fork(1)
@Warmup(iterations = 3, time = 2)
@Measurement(iterations = 4, time = 2)
@OutputTimeUnit(TimeUnit.NANOSECONDS)
@BenchmarkMode(Mode.AverageTime)
public class VarInt32SizeOfBench {
public static final int NUM_OF_VALUES = 1024;

@Param({"1", "2", "4"})
public int numOfBytes;

private int[] numbers;

@Setup
public void setupNumbers() {
Random random = new Random(9387498731984L);
numbers = new int[NUM_OF_VALUES];
final int minValue = numOfBytes == 1 ? 0 : 1 << ((numOfBytes - 1) * 7);
final int maxValue = (1 << (numOfBytes * 7)) - 1;
for (int i = 0; i < NUM_OF_VALUES; i++) {
this.numbers[i] = random.nextInt(minValue, maxValue);
}
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void pbjProtoTools(Blackhole blackhole) {
for (int i = 0; i < NUM_OF_VALUES; i++) blackhole.consume(ProtoWriterTools.sizeOfUnsignedVarInt32(numbers[i]));
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void kafka(Blackhole blackhole) {
for (int i = 0; i < NUM_OF_VALUES; i++) blackhole.consume(kafkaSizeOfUnsignedVarint(numbers[i]));
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void google(Blackhole blackhole) throws IOException {
for (int i = 0; i < NUM_OF_VALUES; i++)
blackhole.consume(com.google.protobuf.CodedOutputStream.computeInt32SizeNoTag(numbers[i]));
}

public static int kafkaSizeOfUnsignedVarint(int value) {
// Protocol buffers varint encoding is variable length, with a minimum of 1 byte
// (for zero). The values themselves are not important. What's important here is
// any leading zero bits are dropped from output. We can use this leading zero
// count w/ fast intrinsic to calc the output length directly.
// Test cases verify this matches the output for loop logic exactly.
// return (38 - leadingZeros) / 7 + leadingZeros / 32;
// The above formula provides the implementation, but the Java encoding is suboptimal
// when we have a narrow range of integers, so we can do better manually
int leadingZeros = Integer.numberOfLeadingZeros(value);
int leadingZerosBelow38DividedBy7 = ((38 - leadingZeros) * 0b10010010010010011) >>> 19;
return leadingZerosBelow38DividedBy7 + (leadingZeros >>> 5);
}

public static void main(String[] args) throws Exception {
Options opt = new OptionsBuilder()
.include(VarInt32SizeOfBench.class.getSimpleName())
.build();

new Runner(opt).run();
}
}
108 changes: 108 additions & 0 deletions ...gration-tests/src/jmh/java/com/hedera/pbj/integration/jmh/varint/VarInt64SizeOfBench.java
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,108 @@
// SPDX-License-Identifier: Apache-2.0
package com.hedera.pbj.integration.jmh.varint;

import com.hedera.pbj.runtime.ProtoWriterTools;
import java.io.IOException;
import java.util.Random;
import java.util.concurrent.TimeUnit;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.CompilerControl;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OperationsPerInvocation;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;
import org.openjdk.jmh.infra.Blackhole;
import org.openjdk.jmh.runner.Runner;
import org.openjdk.jmh.runner.options.Options;
import org.openjdk.jmh.runner.options.OptionsBuilder;

@SuppressWarnings("unused")
@State(Scope.Benchmark)
@Fork(1)
// @Warmup(iterations = 4, time = 2)
// @Measurement(iterations = 5, time = 2)
@Warmup(iterations = 3, time = 2)
@Measurement(iterations = 4, time = 2)
@OutputTimeUnit(TimeUnit.NANOSECONDS)
@BenchmarkMode(Mode.AverageTime)
public class VarInt64SizeOfBench {
public static final int NUM_OF_VALUES = 1024;
/**
* Number of bytes to read at a time (1, 2, 4, or 8). So create inputs with 1 byte siz,e, 2 byte size, 4 byte size,
* and 8 byte size.
*/
@Param({"1", "2", "4", "8"})
public int numOfBytes;

private long[] numbers;

@Setup
public void setupNumbers() {
Random random = new Random(9387498731984L);
numbers = new long[NUM_OF_VALUES];
final long minValue = numOfBytes == 1 ? 0L : 1L << ((numOfBytes - 1) * 7);
final long maxValue = (1L << (numOfBytes * 7)) - 1;
// System.out.println("Generating "+NUM_OF_VALUES+" random numbers between "+min
for (int i = 0; i < NUM_OF_VALUES; i++) {
this.numbers[i] = random.nextLong(minValue, maxValue);
}
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void pbjProtoTools(Blackhole blackhole) throws IOException {
for (int i = 0; i < NUM_OF_VALUES; i++) blackhole.consume(ProtoWriterTools.sizeOfVarInt64(numbers[i]));
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void kafka(Blackhole blackhole) throws IOException {
for (int i = 0; i < NUM_OF_VALUES; i++) blackhole.consume(kafkaSizeOfVarlong(numbers[i]));
}

@Benchmark
@CompilerControl(CompilerControl.Mode.DONT_INLINE)
@OperationsPerInvocation(NUM_OF_VALUES)
public void google(Blackhole blackhole) throws IOException {
for (int i = 0; i < NUM_OF_VALUES; i++)
blackhole.consume(com.google.protobuf.CodedOutputStream.computeInt64SizeNoTag(numbers[i]));
}

public static int kafkaSizeOfVarlong(long value) {
long v = (value << 1) ^ (value >> 63);
int leadingZeros = Long.numberOfLeadingZeros(v);
int leadingZerosBelow70DividedBy7 = ((70 - leadingZeros) * 0b10010010010010011) >>> 19;
return leadingZerosBelow70DividedBy7 + (leadingZeros >>> 6);
}

public static int kafkaSizeOfUnsignedVarint(int value) {
// Protocol buffers varint encoding is variable length, with a minimum of 1 byte
// (for zero). The values themselves are not important. What's important here is
// any leading zero bits are dropped from output. We can use this leading zero
// count w/ fast intrinsic to calc the output length directly.
// Test cases verify this matches the output for loop logic exactly.
// return (38 - leadingZeros) / 7 + leadingZeros / 32;
// The above formula provides the implementation, but the Java encoding is suboptimal
// when we have a narrow range of integers, so we can do better manually
int leadingZeros = Integer.numberOfLeadingZeros(value);
int leadingZerosBelow38DividedBy7 = ((38 - leadingZeros) * 0b10010010010010011) >>> 19;
return leadingZerosBelow38DividedBy7 + (leadingZeros >>> 5);
}

public static void main(String[] args) throws Exception {
Options opt = new OptionsBuilder()
.include(VarInt64SizeOfBench.class.getSimpleName())
.build();

new Runner(opt).run();
}
}
Loading
Loading