Skip to content

Commit

Permalink
[Attempt #2] String overflow hash + data/mem optimization (#356)
Browse files Browse the repository at this point in the history
* Use graal

* Use dynamic cores computer

* Use stream API to cleanup code

* Use max processors

* Use hash to avoid init string

* optimize concurrentmap init

* Smaller hash size

* Avoid checking concurrentmap

* Optimize data type

* string dedup

* Faster write

* Change base

* Remove time

* Use mul instead of div
  • Loading branch information
gnabyl authored Jan 13, 2024
1 parent 45056e0 commit dc49249
Show file tree
Hide file tree
Showing 3 changed files with 83 additions and 77 deletions.
4 changes: 2 additions & 2 deletions calculate_average_gnabyl.sh
Original file line number Diff line number Diff line change
Expand Up @@ -19,5 +19,5 @@
# source "$HOME/.sdkman/bin/sdkman-init.sh"
# sdk use java 21.0.1-graal 1>&2

JAVA_OPTS=""
time java $JAVA_OPTS --class-path target/average-1.0.0-SNAPSHOT.jar dev.morling.onebrc.CalculateAverage_gnabyl
JAVA_OPTS="-XX:+UseStringDeduplication"
java $JAVA_OPTS --class-path target/average-1.0.0-SNAPSHOT.jar dev.morling.onebrc.CalculateAverage_gnabyl
20 changes: 20 additions & 0 deletions prepare_gnabyl.sh
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
#!/bin/bash
#
# Copyright 2023 The original authors
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

# Uncomment below to use sdk
source "$HOME/.sdkman/bin/sdkman-init.sh"
sdk use java 21.0.1-graal 1>&2
136 changes: 61 additions & 75 deletions src/main/java/dev/morling/onebrc/CalculateAverage_gnabyl.java
Original file line number Diff line number Diff line change
Expand Up @@ -16,25 +16,30 @@
package dev.morling.onebrc;

import java.io.IOException;
import java.io.PrintWriter;
import java.io.RandomAccessFile;
import java.nio.MappedByteBuffer;
import java.nio.channels.FileChannel;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.concurrent.CompletableFuture;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ExecutionException;
import java.util.stream.Collectors;

public class CalculateAverage_gnabyl {

private static final String FILE = "./measurements.txt";

private static final int NB_CHUNKS = 8;
private static final int NB_CHUNKS = Runtime.getRuntime().availableProcessors();

private static record Chunk(long start, int bytesCount, MappedByteBuffer mappedByteBuffer) {
private static Map<Integer, String> stationNameMap = new ConcurrentHashMap<>(10000, 0.9f, NB_CHUNKS);

private static record Chunk(int bytesCount, MappedByteBuffer mappedByteBuffer) {
}

private static int reduceSizeToFitLineBreak(FileChannel channel, long startPosition, int startSize)
Expand All @@ -61,26 +66,28 @@ private static int reduceSizeToFitLineBreak(FileChannel channel, long startPosit
return realSize;
}

private static List<Chunk> readChunks(long nbChunks) throws IOException {
private static List<Chunk> readChunks(int nbChunks) throws IOException {
RandomAccessFile file = new RandomAccessFile(FILE, "rw");
List<Chunk> res = new ArrayList<>();
List<Chunk> res = new ArrayList<>(nbChunks);
FileChannel channel = file.getChannel();
long bytesCount = channel.size();
long bytesPerChunk = bytesCount / nbChunks;

// Memory map the file in read-only mode
// TODO: Optimize using threads
long currentPosition = 0;
int startSize;
int realSize;
for (int i = 0; i < nbChunks; i++) {
int startSize = (int) bytesPerChunk;
int realSize = startSize;
startSize = (int) bytesPerChunk;
realSize = startSize;

if (i == nbChunks - 1) {
realSize = (int) (bytesCount - currentPosition);
MappedByteBuffer mappedByteBuffer = channel.map(FileChannel.MapMode.READ_ONLY, currentPosition,
realSize);

res.add(new Chunk(currentPosition, realSize, mappedByteBuffer));
res.add(new Chunk(realSize, mappedByteBuffer));
break;
}

Expand All @@ -90,7 +97,7 @@ private static List<Chunk> readChunks(long nbChunks) throws IOException {
MappedByteBuffer mappedByteBuffer = channel.map(FileChannel.MapMode.READ_ONLY, currentPosition,
realSize);

res.add(new Chunk(currentPosition, realSize, mappedByteBuffer));
res.add(new Chunk(realSize, mappedByteBuffer));
currentPosition += realSize;
}

Expand All @@ -101,32 +108,32 @@ private static List<Chunk> readChunks(long nbChunks) throws IOException {
}

private static class StationData {
private double sum, min, max;
private long count;
private float sum, min, max;
private int count;

public StationData(double value) {
public StationData(float value) {
this.count = 1;
this.sum = value;
this.min = value;
this.max = value;
}

public void update(double value) {
public void update(float value) {
this.count++;
this.sum += value;
this.min = Math.min(this.min, value);
this.max = Math.max(this.max, value);
}

public double getMean() {
public float getMean() {
return sum / count;
}

public double getMin() {
public float getMin() {
return min;
}

public double getMax() {
public float getMax() {
return max;
}

Expand All @@ -139,47 +146,44 @@ public void mergeWith(StationData other) {

}

static double round(double value) {
return Math.round(value * 10.0) / 10.0;
static float round(float value) {
return Math.round(value * 10.0f) * 0.1f;
}

private static class ChunkResult {
private Map<String, StationData> data;
private Map<Integer, StationData> data;

public ChunkResult() {
data = new HashMap<>();
}

public StationData getData(String name) {
return data.get(name);
public StationData getData(int hash) {
return data.get(hash);
}

public void addStation(String name, double value) {
this.data.put(name, new StationData(value));
public void addStation(int hash, float value) {
this.data.put(hash, new StationData(value));
}

public void print() {
var stationNames = new ArrayList<String>(this.data.keySet());
Collections.sort(stationNames);
System.out.print("{");
for (int i = 0; i < stationNames.size() - 1; i++) {
var name = stationNames.get(i);
var stationData = data.get(name);
System.out.printf("%s=%.1f/%.1f/%.1f, ", name, round(stationData.getMin()),
round(stationData.getMean()),
round(stationData.getMax()));
}
var name = stationNames.get(stationNames.size() - 1);
var stationData = data.get(name);
System.out.printf("%s=%.1f/%.1f/%.1f", name, round(stationData.getMin()),
round(stationData.getMean()),
round(stationData.getMax()));
System.out.println("}");
PrintWriter out = new PrintWriter(System.out);
out.println(
this.data.keySet().parallelStream()
.map(hash -> {
var stationData = data.get(hash);
var name = stationNameMap.get(hash);
return String.format("%s=%.1f/%.1f/%.1f", name, round(stationData.getMin()),
round(stationData.getMean()),
round(stationData.getMax()));
})
.sorted((a, b) -> a.split("=")[0].compareTo(b.split("=")[0]))
.collect(Collectors.joining(", ", "{", "}")));
out.flush();
}

public void mergeWith(ChunkResult other) {
for (Map.Entry<String, StationData> entry : other.data.entrySet()) {
String stationName = entry.getKey();
for (Map.Entry<Integer, StationData> entry : other.data.entrySet()) {
int stationName = entry.getKey();
StationData otherStationData = entry.getValue();
StationData thisStationData = this.data.get(stationName);

Expand All @@ -201,16 +205,22 @@ private static ChunkResult processChunk(Chunk chunk) {
chunk.mappedByteBuffer().get(data);

// Process each line
String stationName;
double value;
float value;
int iSplit, iEol;
StationData stationData;
long negative;
int negative;
int hash, prime = 31;
Set<Integer> seenHashes = new HashSet<>(10000, 0.9f);
for (int offset = 0; offset < data.length; offset++) {
// Find station name
hash = 0;
for (iSplit = offset; data[iSplit] != ';'; iSplit++) {
hash = (hash << 5) - hash + (data[iSplit] & 0xFF);
}
if (!seenHashes.contains(hash)) {
seenHashes.add(hash);
stationNameMap.put(hash, new String(data, offset, iSplit - offset, StandardCharsets.UTF_8));
}
stationName = new String(data, offset, iSplit - offset, StandardCharsets.UTF_8);

// Find value
iSplit++;
Expand All @@ -222,7 +232,7 @@ private static ChunkResult processChunk(Chunk chunk) {
continue;
}
if (data[iEol] == '.') {
value = value + (data[iEol + 1] - 48) / 10.0;
value = value + (data[iEol + 1] - 48) * 0.1f;
iEol += 2;
break;
}
Expand All @@ -231,10 +241,10 @@ private static ChunkResult processChunk(Chunk chunk) {
value *= negative;

// Init & count
stationData = result.getData(stationName);
stationData = result.getData(hash);

if (stationData == null) {
result.addStation(stationName, value);
result.addStation(hash, value);
}
else {
stationData.update(value);
Expand All @@ -247,32 +257,8 @@ private static ChunkResult processChunk(Chunk chunk) {
}

private static ChunkResult processAllChunks(List<Chunk> chunks) throws InterruptedException, ExecutionException {
// var globalRes = new ChunkResult();
// for (var chunk : chunks) {
// var chunkRes = processChunk(chunk);
// globalRes.mergeWith(chunkRes);
// }
// return globalRes;

List<CompletableFuture<ChunkResult>> computeTasks = new ArrayList<>();

for (Chunk chunk : chunks) {
computeTasks.add(CompletableFuture.supplyAsync(() -> processChunk(chunk)));
}

ChunkResult globalRes = null;

for (CompletableFuture<ChunkResult> completedTask : computeTasks) {
ChunkResult chunkRes = completedTask.get();
if (globalRes == null) {
globalRes = completedTask.get();
}
else {
globalRes.mergeWith(chunkRes);
}
}

return globalRes;
return chunks.parallelStream().map(CalculateAverage_gnabyl::processChunk).collect(ChunkResult::new,
ChunkResult::mergeWith, ChunkResult::mergeWith);
}

public static void main(String[] args) throws IOException, InterruptedException, ExecutionException {
Expand Down

0 comments on commit dc49249

Please sign in to comment.