perf: vectorize topN native engine #19353
There are no files selected for viewing
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
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,218 @@ | ||
| /* | ||
| * Licensed to the Apache Software Foundation (ASF) under one | ||
| * or more contributor license agreements. See the NOTICE file | ||
| * distributed with this work for additional information | ||
| * regarding copyright ownership. The ASF licenses this file | ||
| * to you 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. | ||
| */ | ||
|
|
||
| package org.apache.druid.query.groupby.epinephelinae; | ||
|
|
||
| import it.unimi.dsi.fastutil.Hash; | ||
| import it.unimi.dsi.fastutil.objects.Object2IntMap; | ||
| import it.unimi.dsi.fastutil.objects.Object2IntOpenCustomHashMap; | ||
| import org.apache.datasketches.memory.Memory; | ||
| import org.apache.druid.java.util.common.ISE; | ||
| import org.apache.druid.java.util.common.parsers.CloseableIterator; | ||
| import org.apache.druid.query.aggregation.AggregatorAdapters; | ||
| import org.apache.druid.query.groupby.epinephelinae.collection.MemoryPointer; | ||
|
|
||
| import java.nio.ByteBuffer; | ||
| import java.nio.ByteOrder; | ||
| import java.util.Arrays; | ||
| import java.util.Iterator; | ||
|
|
||
| /** | ||
| * On-heap {@link VectorGrouper} that grows aggregator state on demand, up to maximum limit of 2GB. | ||
| * | ||
| * Vectorized analogue of {@link org.apache.druid.query.topn.BaseTopNAlgorithm}'s | ||
| * {@code runWithCardinalityUnknown} path: used when dimension cardinality is unknown (numeric columns, | ||
| * non-dict-encoded string virtual columns) or when a dict-encoded string column's cardinality exceeds | ||
| * the processing buffer. Memory footprint is on-heap and grows with the distinct-key count — matching | ||
| * the non-vectorized path's memory profile for the same queries. | ||
| */ | ||
| public class HeapVectorGrouper implements VectorGrouper | ||
| { | ||
| private static final Hash.Strategy<byte[]> BYTE_ARRAY_HASH_STRATEGY = new Hash.Strategy<byte[]>() | ||
| { | ||
| @Override | ||
| public int hashCode(byte[] o) | ||
| { | ||
| return Arrays.hashCode(o); | ||
| } | ||
|
|
||
| @Override | ||
| public boolean equals(byte[] a, byte[] b) | ||
| { | ||
| return Arrays.equals(a, b); | ||
| } | ||
| }; | ||
|
|
||
| private static final int MIN_INITIAL_STATE_BUFFER_SIZE = 4096; | ||
|
|
||
| private final AggregatorAdapters aggregators; | ||
| private final int keySize; | ||
| private final int aggStateSize; | ||
| private final Object2IntOpenCustomHashMap<byte[]> keyToOffset; | ||
|
|
||
| private boolean initialized; | ||
| private ByteBuffer aggStateBuffer; | ||
| private int aggStateEnd; | ||
|
|
||
| private int[] vAggregationPositions; | ||
| private int[] vAggregationRows; | ||
| private byte[] keyScratch; | ||
|
|
||
| public HeapVectorGrouper(final AggregatorAdapters aggregators, final int keySize) | ||
| { | ||
| this.aggregators = aggregators; | ||
| this.keySize = keySize; | ||
| this.aggStateSize = aggregators.spaceNeeded(); | ||
| this.keyToOffset = new Object2IntOpenCustomHashMap<>(BYTE_ARRAY_HASH_STRATEGY); | ||
| this.keyToOffset.defaultReturnValue(-1); | ||
| } | ||
|
|
||
| @Override | ||
| public void initVectorized(final int maxVectorSize) | ||
| { | ||
| if (initialized) { | ||
| if (vAggregationPositions.length != maxVectorSize) { | ||
| throw new ISE( | ||
| "initVectorized called with different maxVectorSize (existing=%d, new=%d)", | ||
| vAggregationPositions.length, | ||
| maxVectorSize | ||
| ); | ||
| } | ||
| return; | ||
| } | ||
| this.aggStateBuffer = ByteBuffer.allocate(MIN_INITIAL_STATE_BUFFER_SIZE); | ||
| this.vAggregationPositions = new int[maxVectorSize]; | ||
| this.vAggregationRows = new int[maxVectorSize]; | ||
| this.keyScratch = new byte[keySize]; | ||
| this.aggStateEnd = 0; | ||
| this.initialized = true; | ||
| } | ||
|
|
||
| /** | ||
| * Contract: keys for rows [startRow, endRow) must be packed contiguously at {@code keySpace[0 .. | ||
| * numRows * keySize)}; {@code startRow}/{@code endRow} are source-vector indices used to look up aggregator | ||
| * input values. | ||
| */ | ||
| @Override | ||
| public AggregateResult aggregateVector(final Memory keySpace, final int startRow, final int endRow) | ||
| { | ||
| final int numRows = endRow - startRow; | ||
|
|
||
| for (int i = 0; i < numRows; i++) { | ||
| keySpace.getByteArray((long) i * keySize, keyScratch, 0, keySize); | ||
| int offset = keyToOffset.getInt(keyScratch); | ||
| if (offset == -1) { | ||
| if ((long) aggStateEnd + aggStateSize > aggStateBuffer.capacity()) { | ||
| growBuffer((long) aggStateEnd + aggStateSize); | ||
| } | ||
| offset = aggStateEnd; | ||
| final byte[] keyCopy = Arrays.copyOf(keyScratch, keySize); | ||
| keyToOffset.put(keyCopy, offset); | ||
| aggregators.init(aggStateBuffer, offset); | ||
| aggStateEnd += aggStateSize; | ||
| } | ||
| vAggregationPositions[i] = offset; | ||
| } | ||
|
|
||
| aggregators.aggregateVector( | ||
| aggStateBuffer, | ||
| numRows, | ||
| vAggregationPositions, | ||
| Groupers.writeAggregationRows(vAggregationRows, startRow, endRow) | ||
| ); | ||
|
|
||
| return AggregateResult.ok(); | ||
| } | ||
|
|
||
| private void growBuffer(final long neededCapacity) | ||
| { | ||
Check failureCode scanning / CodeQL Comparison of narrow type with wide type in loop condition High |
||
| if (neededCapacity > Integer.MAX_VALUE) { | ||
| throw new ISE("Aggregator state exceeds 2 GB; cardinality too high for HeapVectorGrouper"); | ||
| } | ||
| int newCapacity = aggStateBuffer.capacity(); | ||
| while (newCapacity < neededCapacity) { | ||
|
|
||
| final long doubled = (long) newCapacity * 2; | ||
| newCapacity = doubled > Integer.MAX_VALUE ? Integer.MAX_VALUE : (int) doubled; | ||
| } | ||
|
|
||
| final ByteBuffer oldBuffer = aggStateBuffer; | ||
| final ByteBuffer newBuffer = ByteBuffer.allocate(newCapacity); | ||
| oldBuffer.position(0); | ||
| oldBuffer.limit(aggStateEnd); | ||
| newBuffer.put(oldBuffer); | ||
|
|
||
| for (int pos = 0; pos < aggStateEnd; pos += aggStateSize) { | ||
| aggregators.relocate(pos, pos, oldBuffer, newBuffer); | ||
| } | ||
|
|
||
| this.aggStateBuffer = newBuffer; | ||
| } | ||
|
|
||
| @Override | ||
| public void reset() | ||
| { | ||
| aggregators.reset(); | ||
| keyToOffset.clear(); | ||
| aggStateEnd = 0; | ||
| } | ||
|
|
||
| @Override | ||
| public void close() | ||
| { | ||
| reset(); | ||
| } | ||
|
|
||
| @Override | ||
| public CloseableIterator<Grouper.Entry<MemoryPointer>> iterator() | ||
| { | ||
| final Iterator<Object2IntMap.Entry<byte[]>> mapIter = | ||
| keyToOffset.object2IntEntrySet().fastIterator(); | ||
|
|
||
| return new CloseableIterator<>() | ||
| { | ||
| final ReusableEntry<MemoryPointer> reusableEntry = | ||
| new ReusableEntry<>(new MemoryPointer(), new Object[aggregators.size()]); | ||
|
|
||
| @Override | ||
| public boolean hasNext() | ||
| { | ||
| return mapIter.hasNext(); | ||
| } | ||
|
|
||
| @Override | ||
| public Grouper.Entry<MemoryPointer> next() | ||
| { | ||
| final Object2IntMap.Entry<byte[]> mapEntry = mapIter.next(); | ||
| reusableEntry.getKey().set(Memory.wrap(mapEntry.getKey(), ByteOrder.nativeOrder()), 0); | ||
|
|
||
| final int position = mapEntry.getIntValue(); | ||
| for (int i = 0; i < aggregators.size(); i++) { | ||
| reusableEntry.getValues()[i] = aggregators.get(aggStateBuffer, position, i); | ||
| } | ||
| return reusableEntry; | ||
| } | ||
|
|
||
| @Override | ||
| public void close() | ||
| { | ||
| // Nothing to close. | ||
| } | ||
| }; | ||
| } | ||
| } | ||
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
Oops, something went wrong.
Oops, something went wrong.
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
probably want to make this limit configurable