benwtrent commented on code in PR #14131: URL: https://github.com/apache/lucene/pull/14131#discussion_r1914748083
########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsWriter.java: ########## @@ -0,0 +1,402 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceIndexParams; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraIndexParams; +import com.nvidia.cuvs.CagraIndexParams.CagraGraphBuildAlgo; +import com.nvidia.cuvs.CuVSResources; +import java.io.ByteArrayOutputStream; +import java.io.File; +import java.io.IOException; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.LinkedHashMap; +import java.util.List; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnFieldVectorsWriter; +import org.apache.lucene.codecs.KnnVectorsWriter; +import org.apache.lucene.index.FieldInfo; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.MergeState; +import org.apache.lucene.index.SegmentWriteState; +import org.apache.lucene.index.Sorter.DocMap; +import org.apache.lucene.store.IndexOutput; +import org.apache.lucene.util.IOUtils; +import org.apache.lucene.util.SuppressForbidden; + +public class CuVSVectorsWriter extends KnnVectorsWriter { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + private List<CagraFieldVectorsWriter> fieldVectorWriters = new ArrayList<>(); + private IndexOutput cuVSIndex = null; + private SegmentWriteState segmentWriteState = null; + private String cuVSDataFilename = null; + + private CagraIndex cagraIndex; + private CagraIndex cagraIndexForHnsw; + + private int cuvsWriterThreads; + private int intGraphDegree; + private int graphDegree; + private MergeStrategy mergeStrategy; + private CuVSResources resources; + + public enum MergeStrategy { + TRIVIAL_MERGE, + NON_TRIVIAL_MERGE + }; + + public CuVSVectorsWriter( + SegmentWriteState state, + int cuvsWriterThreads, + int intGraphDegree, + int graphDegree, + MergeStrategy mergeStrategy, + CuVSResources resources) + throws IOException { + super(); + this.segmentWriteState = state; + this.mergeStrategy = mergeStrategy; + this.cuvsWriterThreads = cuvsWriterThreads; + this.intGraphDegree = intGraphDegree; + this.graphDegree = graphDegree; + this.resources = resources; + + cuVSDataFilename = + IndexFileNames.segmentFileName( + this.segmentWriteState.segmentInfo.name, + this.segmentWriteState.segmentSuffix, + CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + } + + @Override + public long ramBytesUsed() { + return 0; + } + + @Override + public void close() throws IOException { + IOUtils.close(cuVSIndex); + cuVSIndex = null; + fieldVectorWriters.clear(); + fieldVectorWriters = null; + } + + @Override + public KnnFieldVectorsWriter<?> addField(FieldInfo fieldInfo) throws IOException { + CagraFieldVectorsWriter cagraFieldVectorWriter = new CagraFieldVectorsWriter(fieldInfo); + fieldVectorWriters.add(cagraFieldVectorWriter); + return cagraFieldVectorWriter; + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for Cagra's serialization") + private byte[] createCagraIndex(float[][] vectors, List<Integer> mapping) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndex = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = + File.createTempFile( + "tmpindex", "cag"); // TODO: Should we make this a file with random names? + cagraIndex.serialize(baos, tmpFile); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for BruteForce's serialization") + private byte[] createBruteForceIndex(float[][] vectors) throws Throwable { + BruteForceIndexParams indexParams = + new BruteForceIndexParams.Builder() + .withNumWriterThreads(32) // TODO: Make this configurable later. + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + BruteForceIndex index = + new BruteForceIndex.Builder(resources) + .withIndexParams(indexParams) + .withDataset(vectors) + .build(); + + // log.info("Indexing done: " + System.currentTimeMillis()); + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + index.serialize(baos); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for HNSW's serialization") + private byte[] createHnswIndex(float[][] vectors) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndexForHnsw = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = File.createTempFile("tmpindex", "hnsw"); + cagraIndexForHnsw.serializeToHNSW(baos, tmpFile); + return baos.toByteArray(); + } + + @SuppressWarnings({"resource", "rawtypes", "unchecked"}) + @Override + public void flush(int maxDoc, DocMap sortMap) throws IOException { + cuVSIndex = + this.segmentWriteState.directory.createOutput( + cuVSDataFilename, this.segmentWriteState.context); + CodecUtil.writeIndexHeader( + cuVSIndex, + CuVSVectorsFormat.VECTOR_DATA_CODEC_NAME, + CuVSVectorsFormat.VERSION_CURRENT, + this.segmentWriteState.segmentInfo.getId(), + this.segmentWriteState.segmentSuffix); + + CuVSSegmentFile cuVSFile = new CuVSSegmentFile(new SegmentOutputStream(cuVSIndex, 100000)); + + LinkedHashMap<String, Integer> metaMap = new LinkedHashMap<String, Integer>(); + + for (CagraFieldVectorsWriter field : fieldVectorWriters) { + // long start = System.currentTimeMillis(); + + byte[] cagraIndexBytes = null; + byte[] bruteForceIndexBytes = null; + byte[] hnswIndexBytes = null; + try { + // log.info("Starting CAGRA indexing, space remaining: " + new File("/").getFreeSpace()); + // log.info("Starting CAGRA indexing, docs: " + field.vectors.size()); + + float vectors[][] = new float[field.vectors.size()][field.vectors.get(0).length]; + for (int i = 0; i < vectors.length; i++) { + for (int j = 0; j < vectors[i].length; j++) { + vectors[i][j] = field.vectors.get(i)[j]; + } + } + + cagraIndexBytes = createCagraIndex(vectors, new ArrayList<Integer>(field.vectors.keySet())); + bruteForceIndexBytes = createBruteForceIndex(vectors); + hnswIndexBytes = createHnswIndex(vectors); Review Comment: From what I can tell, you load all the vectors onto heap. This is frankly untenable in production. The amount of GC & JVM heap would be enormous on medium size indices (10M+). ########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsWriter.java: ########## @@ -0,0 +1,402 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceIndexParams; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraIndexParams; +import com.nvidia.cuvs.CagraIndexParams.CagraGraphBuildAlgo; +import com.nvidia.cuvs.CuVSResources; +import java.io.ByteArrayOutputStream; +import java.io.File; +import java.io.IOException; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.LinkedHashMap; +import java.util.List; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnFieldVectorsWriter; +import org.apache.lucene.codecs.KnnVectorsWriter; +import org.apache.lucene.index.FieldInfo; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.MergeState; +import org.apache.lucene.index.SegmentWriteState; +import org.apache.lucene.index.Sorter.DocMap; +import org.apache.lucene.store.IndexOutput; +import org.apache.lucene.util.IOUtils; +import org.apache.lucene.util.SuppressForbidden; + +public class CuVSVectorsWriter extends KnnVectorsWriter { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + private List<CagraFieldVectorsWriter> fieldVectorWriters = new ArrayList<>(); + private IndexOutput cuVSIndex = null; + private SegmentWriteState segmentWriteState = null; + private String cuVSDataFilename = null; + + private CagraIndex cagraIndex; + private CagraIndex cagraIndexForHnsw; + + private int cuvsWriterThreads; + private int intGraphDegree; + private int graphDegree; + private MergeStrategy mergeStrategy; + private CuVSResources resources; + + public enum MergeStrategy { + TRIVIAL_MERGE, + NON_TRIVIAL_MERGE + }; + + public CuVSVectorsWriter( + SegmentWriteState state, + int cuvsWriterThreads, + int intGraphDegree, + int graphDegree, + MergeStrategy mergeStrategy, + CuVSResources resources) + throws IOException { + super(); + this.segmentWriteState = state; + this.mergeStrategy = mergeStrategy; + this.cuvsWriterThreads = cuvsWriterThreads; + this.intGraphDegree = intGraphDegree; + this.graphDegree = graphDegree; + this.resources = resources; + + cuVSDataFilename = + IndexFileNames.segmentFileName( + this.segmentWriteState.segmentInfo.name, + this.segmentWriteState.segmentSuffix, + CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + } + + @Override + public long ramBytesUsed() { + return 0; + } + + @Override + public void close() throws IOException { + IOUtils.close(cuVSIndex); + cuVSIndex = null; + fieldVectorWriters.clear(); + fieldVectorWriters = null; + } + + @Override + public KnnFieldVectorsWriter<?> addField(FieldInfo fieldInfo) throws IOException { + CagraFieldVectorsWriter cagraFieldVectorWriter = new CagraFieldVectorsWriter(fieldInfo); + fieldVectorWriters.add(cagraFieldVectorWriter); + return cagraFieldVectorWriter; + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for Cagra's serialization") + private byte[] createCagraIndex(float[][] vectors, List<Integer> mapping) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndex = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = + File.createTempFile( + "tmpindex", "cag"); // TODO: Should we make this a file with random names? + cagraIndex.serialize(baos, tmpFile); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for BruteForce's serialization") + private byte[] createBruteForceIndex(float[][] vectors) throws Throwable { + BruteForceIndexParams indexParams = + new BruteForceIndexParams.Builder() + .withNumWriterThreads(32) // TODO: Make this configurable later. + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + BruteForceIndex index = + new BruteForceIndex.Builder(resources) + .withIndexParams(indexParams) + .withDataset(vectors) + .build(); + + // log.info("Indexing done: " + System.currentTimeMillis()); + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + index.serialize(baos); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for HNSW's serialization") + private byte[] createHnswIndex(float[][] vectors) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndexForHnsw = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); Review Comment: The cagra index builder should instead accept a file handle or something. Having to feed it on-heap vectors is not good. ########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsWriter.java: ########## @@ -0,0 +1,402 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceIndexParams; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraIndexParams; +import com.nvidia.cuvs.CagraIndexParams.CagraGraphBuildAlgo; +import com.nvidia.cuvs.CuVSResources; +import java.io.ByteArrayOutputStream; +import java.io.File; +import java.io.IOException; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.LinkedHashMap; +import java.util.List; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnFieldVectorsWriter; +import org.apache.lucene.codecs.KnnVectorsWriter; +import org.apache.lucene.index.FieldInfo; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.MergeState; +import org.apache.lucene.index.SegmentWriteState; +import org.apache.lucene.index.Sorter.DocMap; +import org.apache.lucene.store.IndexOutput; +import org.apache.lucene.util.IOUtils; +import org.apache.lucene.util.SuppressForbidden; + +public class CuVSVectorsWriter extends KnnVectorsWriter { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + private List<CagraFieldVectorsWriter> fieldVectorWriters = new ArrayList<>(); + private IndexOutput cuVSIndex = null; + private SegmentWriteState segmentWriteState = null; + private String cuVSDataFilename = null; + + private CagraIndex cagraIndex; + private CagraIndex cagraIndexForHnsw; + + private int cuvsWriterThreads; + private int intGraphDegree; + private int graphDegree; + private MergeStrategy mergeStrategy; + private CuVSResources resources; + + public enum MergeStrategy { + TRIVIAL_MERGE, + NON_TRIVIAL_MERGE + }; + + public CuVSVectorsWriter( + SegmentWriteState state, + int cuvsWriterThreads, + int intGraphDegree, + int graphDegree, + MergeStrategy mergeStrategy, + CuVSResources resources) + throws IOException { + super(); + this.segmentWriteState = state; + this.mergeStrategy = mergeStrategy; + this.cuvsWriterThreads = cuvsWriterThreads; + this.intGraphDegree = intGraphDegree; + this.graphDegree = graphDegree; + this.resources = resources; + + cuVSDataFilename = + IndexFileNames.segmentFileName( + this.segmentWriteState.segmentInfo.name, + this.segmentWriteState.segmentSuffix, + CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + } + + @Override + public long ramBytesUsed() { + return 0; + } + + @Override + public void close() throws IOException { + IOUtils.close(cuVSIndex); + cuVSIndex = null; + fieldVectorWriters.clear(); + fieldVectorWriters = null; + } + + @Override + public KnnFieldVectorsWriter<?> addField(FieldInfo fieldInfo) throws IOException { + CagraFieldVectorsWriter cagraFieldVectorWriter = new CagraFieldVectorsWriter(fieldInfo); + fieldVectorWriters.add(cagraFieldVectorWriter); + return cagraFieldVectorWriter; + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for Cagra's serialization") + private byte[] createCagraIndex(float[][] vectors, List<Integer> mapping) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndex = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = + File.createTempFile( + "tmpindex", "cag"); // TODO: Should we make this a file with random names? + cagraIndex.serialize(baos, tmpFile); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for BruteForce's serialization") + private byte[] createBruteForceIndex(float[][] vectors) throws Throwable { + BruteForceIndexParams indexParams = + new BruteForceIndexParams.Builder() + .withNumWriterThreads(32) // TODO: Make this configurable later. + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + BruteForceIndex index = + new BruteForceIndex.Builder(resources) + .withIndexParams(indexParams) + .withDataset(vectors) + .build(); + + // log.info("Indexing done: " + System.currentTimeMillis()); + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + index.serialize(baos); + return baos.toByteArray(); + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for HNSW's serialization") + private byte[] createHnswIndex(float[][] vectors) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndexForHnsw = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = File.createTempFile("tmpindex", "hnsw"); + cagraIndexForHnsw.serializeToHNSW(baos, tmpFile); + return baos.toByteArray(); + } + + @SuppressWarnings({"resource", "rawtypes", "unchecked"}) + @Override + public void flush(int maxDoc, DocMap sortMap) throws IOException { + cuVSIndex = + this.segmentWriteState.directory.createOutput( + cuVSDataFilename, this.segmentWriteState.context); + CodecUtil.writeIndexHeader( + cuVSIndex, + CuVSVectorsFormat.VECTOR_DATA_CODEC_NAME, + CuVSVectorsFormat.VERSION_CURRENT, + this.segmentWriteState.segmentInfo.getId(), + this.segmentWriteState.segmentSuffix); + + CuVSSegmentFile cuVSFile = new CuVSSegmentFile(new SegmentOutputStream(cuVSIndex, 100000)); Review Comment: Why not just use all the built in inputs/outputs? Seems weird to have a unique buffered output. ########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsWriter.java: ########## @@ -0,0 +1,402 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceIndexParams; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraIndexParams; +import com.nvidia.cuvs.CagraIndexParams.CagraGraphBuildAlgo; +import com.nvidia.cuvs.CuVSResources; +import java.io.ByteArrayOutputStream; +import java.io.File; +import java.io.IOException; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.LinkedHashMap; +import java.util.List; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnFieldVectorsWriter; +import org.apache.lucene.codecs.KnnVectorsWriter; +import org.apache.lucene.index.FieldInfo; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.MergeState; +import org.apache.lucene.index.SegmentWriteState; +import org.apache.lucene.index.Sorter.DocMap; +import org.apache.lucene.store.IndexOutput; +import org.apache.lucene.util.IOUtils; +import org.apache.lucene.util.SuppressForbidden; + +public class CuVSVectorsWriter extends KnnVectorsWriter { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + private List<CagraFieldVectorsWriter> fieldVectorWriters = new ArrayList<>(); + private IndexOutput cuVSIndex = null; + private SegmentWriteState segmentWriteState = null; + private String cuVSDataFilename = null; + + private CagraIndex cagraIndex; + private CagraIndex cagraIndexForHnsw; + + private int cuvsWriterThreads; + private int intGraphDegree; + private int graphDegree; + private MergeStrategy mergeStrategy; + private CuVSResources resources; + + public enum MergeStrategy { + TRIVIAL_MERGE, + NON_TRIVIAL_MERGE + }; + + public CuVSVectorsWriter( + SegmentWriteState state, + int cuvsWriterThreads, + int intGraphDegree, + int graphDegree, + MergeStrategy mergeStrategy, + CuVSResources resources) + throws IOException { + super(); + this.segmentWriteState = state; + this.mergeStrategy = mergeStrategy; + this.cuvsWriterThreads = cuvsWriterThreads; + this.intGraphDegree = intGraphDegree; + this.graphDegree = graphDegree; + this.resources = resources; + + cuVSDataFilename = + IndexFileNames.segmentFileName( + this.segmentWriteState.segmentInfo.name, + this.segmentWriteState.segmentSuffix, + CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + } + + @Override + public long ramBytesUsed() { + return 0; + } + + @Override + public void close() throws IOException { + IOUtils.close(cuVSIndex); + cuVSIndex = null; + fieldVectorWriters.clear(); + fieldVectorWriters = null; + } + + @Override + public KnnFieldVectorsWriter<?> addField(FieldInfo fieldInfo) throws IOException { + CagraFieldVectorsWriter cagraFieldVectorWriter = new CagraFieldVectorsWriter(fieldInfo); + fieldVectorWriters.add(cagraFieldVectorWriter); + return cagraFieldVectorWriter; + } + + @SuppressForbidden(reason = "A temporary java.util.File is needed for Cagra's serialization") + private byte[] createCagraIndex(float[][] vectors, List<Integer> mapping) throws Throwable { + CagraIndexParams indexParams = + new CagraIndexParams.Builder(resources) + .withNumWriterThreads(cuvsWriterThreads) + .withIntermediateGraphDegree(intGraphDegree) + .withGraphDegree(graphDegree) + .withCagraGraphBuildAlgo(CagraGraphBuildAlgo.NN_DESCENT) + .build(); + + // log.info("Indexing started: " + System.currentTimeMillis()); + cagraIndex = + new CagraIndex.Builder(resources).withDataset(vectors).withIndexParams(indexParams).build(); + // log.info("Indexing done: " + System.currentTimeMillis() + "ms, documents: " + + // vectors.length); + + ByteArrayOutputStream baos = new ByteArrayOutputStream(); + File tmpFile = + File.createTempFile( + "tmpindex", "cag"); // TODO: Should we make this a file with random names? Review Comment: This tmp file name basically means you cannot have different fields with different settings running at the same time? Seems like a very bad idea. ########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsReader.java: ########## @@ -0,0 +1,340 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceQuery; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraQuery; +import com.nvidia.cuvs.CagraSearchParams; +import com.nvidia.cuvs.CuVSResources; +import com.nvidia.cuvs.HnswIndex; +import com.nvidia.cuvs.HnswIndexParams; +import java.io.ByteArrayInputStream; +import java.io.ByteArrayOutputStream; +import java.io.IOException; +import java.lang.StackWalker.StackFrame; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.Map.Entry; +import java.util.stream.Collectors; +import java.util.stream.Stream; +import java.util.zip.ZipEntry; +import java.util.zip.ZipInputStream; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnVectorsReader; +import org.apache.lucene.index.ByteVectorValues; +import org.apache.lucene.index.FloatVectorValues; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.SegmentReadState; +import org.apache.lucene.search.KnnCollector; +import org.apache.lucene.search.TopKnnCollector; +import org.apache.lucene.store.IndexInput; +import org.apache.lucene.util.Bits; +import org.apache.lucene.util.FixedBitSet; +import org.apache.lucene.util.IOUtils; + +public class CuVSVectorsReader extends KnnVectorsReader { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + IndexInput vectorDataReader = null; + public String fileName = null; + public byte[] indexFileBytes; + public int[] docIds; + public float[] vectors; + public SegmentReadState segmentState = null; + public int indexFilePayloadSize = 0; + public long initialFilePointerLoc = 0; + public SegmentInputStream segmentInputStream; + + // Field to List of Indexes + public Map<String, List<CuVSIndex>> cuvsIndexes; + + private CuVSResources resources; + + public CuVSVectorsReader(SegmentReadState state, CuVSResources resources) throws Throwable { + + segmentState = state; + this.resources = resources; + + fileName = + IndexFileNames.segmentFileName( + state.segmentInfo.name, state.segmentSuffix, CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + + vectorDataReader = segmentState.directory.openInput(fileName, segmentState.context); + CodecUtil.readIndexHeader(vectorDataReader); + + initialFilePointerLoc = vectorDataReader.getFilePointer(); + indexFilePayloadSize = + (int) vectorDataReader.length() + - (int) initialFilePointerLoc; // vectorMetaReader.readInt(); + segmentInputStream = + new SegmentInputStream(vectorDataReader, indexFilePayloadSize, initialFilePointerLoc); + // log.info("payloadSize: " + indexFilePayloadSize); + // log.info("initialFilePointerLoc: " + initialFilePointerLoc); + + List<StackFrame> stackTrace = StackWalker.getInstance().walk(this::getStackTrace); + + boolean isMergeCase = false; + for (StackFrame s : stackTrace) { + if (s.toString().startsWith("org.apache.lucene.index.IndexWriter.merge")) { + isMergeCase = true; + // log.info("Reader opening on merge call"); + break; Review Comment: You should instead use `getMergeInstance` which allows you do set merge settings, or whatever for a given reader. ########## lucene/sandbox/src/java/org/apache/lucene/sandbox/vectorsearch/CuVSVectorsReader.java: ########## @@ -0,0 +1,340 @@ +/* + * 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.lucene.sandbox.vectorsearch; + +import com.nvidia.cuvs.BruteForceIndex; +import com.nvidia.cuvs.BruteForceQuery; +import com.nvidia.cuvs.CagraIndex; +import com.nvidia.cuvs.CagraQuery; +import com.nvidia.cuvs.CagraSearchParams; +import com.nvidia.cuvs.CuVSResources; +import com.nvidia.cuvs.HnswIndex; +import com.nvidia.cuvs.HnswIndexParams; +import java.io.ByteArrayInputStream; +import java.io.ByteArrayOutputStream; +import java.io.IOException; +import java.lang.StackWalker.StackFrame; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.Map.Entry; +import java.util.stream.Collectors; +import java.util.stream.Stream; +import java.util.zip.ZipEntry; +import java.util.zip.ZipInputStream; +import org.apache.commons.lang3.SerializationUtils; +import org.apache.lucene.codecs.CodecUtil; +import org.apache.lucene.codecs.KnnVectorsReader; +import org.apache.lucene.index.ByteVectorValues; +import org.apache.lucene.index.FloatVectorValues; +import org.apache.lucene.index.IndexFileNames; +import org.apache.lucene.index.SegmentReadState; +import org.apache.lucene.search.KnnCollector; +import org.apache.lucene.search.TopKnnCollector; +import org.apache.lucene.store.IndexInput; +import org.apache.lucene.util.Bits; +import org.apache.lucene.util.FixedBitSet; +import org.apache.lucene.util.IOUtils; + +public class CuVSVectorsReader extends KnnVectorsReader { + + // protected Logger log = Logger.getLogger(getClass().getName()); + + IndexInput vectorDataReader = null; + public String fileName = null; + public byte[] indexFileBytes; + public int[] docIds; + public float[] vectors; + public SegmentReadState segmentState = null; + public int indexFilePayloadSize = 0; + public long initialFilePointerLoc = 0; + public SegmentInputStream segmentInputStream; + + // Field to List of Indexes + public Map<String, List<CuVSIndex>> cuvsIndexes; + + private CuVSResources resources; + + public CuVSVectorsReader(SegmentReadState state, CuVSResources resources) throws Throwable { + + segmentState = state; + this.resources = resources; + + fileName = + IndexFileNames.segmentFileName( + state.segmentInfo.name, state.segmentSuffix, CuVSVectorsFormat.VECTOR_DATA_EXTENSION); + + vectorDataReader = segmentState.directory.openInput(fileName, segmentState.context); + CodecUtil.readIndexHeader(vectorDataReader); + + initialFilePointerLoc = vectorDataReader.getFilePointer(); + indexFilePayloadSize = + (int) vectorDataReader.length() + - (int) initialFilePointerLoc; // vectorMetaReader.readInt(); + segmentInputStream = + new SegmentInputStream(vectorDataReader, indexFilePayloadSize, initialFilePointerLoc); + // log.info("payloadSize: " + indexFilePayloadSize); + // log.info("initialFilePointerLoc: " + initialFilePointerLoc); + + List<StackFrame> stackTrace = StackWalker.getInstance().walk(this::getStackTrace); + + boolean isMergeCase = false; + for (StackFrame s : stackTrace) { + if (s.toString().startsWith("org.apache.lucene.index.IndexWriter.merge")) { + isMergeCase = true; + // log.info("Reader opening on merge call"); + break; + } + } + + /*log.info( + "Source of this segment " + + segmentState.segmentSuffix + + " is " + + segmentState.segmentInfo.getDiagnostics().get(IndexWriter.SOURCE)); + log.info("Loading for " + segmentState.segmentInfo.name + ", mergeCase? " + isMergeCase); + log.info("Not the merge case, hence loading for " + segmentState.segmentInfo.name);*/ + this.cuvsIndexes = loadCuVSIndex(getIndexInputStream(), isMergeCase); + } + + @SuppressWarnings({"unchecked"}) + private Map<String, List<CuVSIndex>> loadCuVSIndex(ZipInputStream zis, boolean isMergeCase) + throws Throwable { + Map<String, List<CuVSIndex>> ret = new HashMap<String, List<CuVSIndex>>(); + Map<String, CagraIndex> cagraIndexes = new HashMap<String, CagraIndex>(); + Map<String, BruteForceIndex> bruteforceIndexes = new HashMap<String, BruteForceIndex>(); + Map<String, HnswIndex> hnswIndexes = new HashMap<String, HnswIndex>(); + Map<String, List<Integer>> mappings = new HashMap<String, List<Integer>>(); + Map<String, List<float[]>> vectors = new HashMap<String, List<float[]>>(); Review Comment: From what I can tell, every time this reader is opened, you load everything on heap and THEN build the cuvsindex by scratch? Why not serialize the cuvindex itself? Doesn't it have a file format or something? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
