Skip to content

Commit f8b7217

Browse files
szybiaszybia
committed
Merge remote-tracking branch 'upstream/main' into index-pipeline-tracking
* upstream/main: (91 commits) Mute org.elasticsearch.packaging.test.DockerTests test130JavaHasCorrectOwnership elastic#131369 Add exception logging when interrupted (elastic#131153) Mute org.elasticsearch.packaging.test.DockerTests test140CgroupOsStatsAreAvailable elastic#131372 Mute org.elasticsearch.packaging.test.DockerTests test070BindMountCustomPathConfAndJvmOptions elastic#131366 Mute org.elasticsearch.xpack.test.rest.XPackRestIT test {p0=ml/delete_expired_data/Test delete expired data with body parameters} elastic#131364 Mute org.elasticsearch.xpack.esql.vector.VectorSimilarityFunctionsIT testSimilarityBetweenConstantVectorAndField {functionName=v_cosine similarityFunction=COSINE} elastic#131363 Mute org.elasticsearch.xpack.esql.vector.VectorSimilarityFunctionsIT testDifferentDimensions {functionName=v_cosine similarityFunction=COSINE} elastic#131362 Mute org.elasticsearch.xpack.esql.vector.VectorSimilarityFunctionsIT testSimilarityBetweenConstantVectors {functionName=v_cosine similarityFunction=COSINE} elastic#131361 Check SCORE_FUNCTION capability in VerifierTests (elastic#131352) Replace deprecated routingTable table call in tests (elastic#131005) [DOCS] Remove misused applies_to tag (elastic#131349) Adj ivf postings list building (elastic#130843) [Transform] Read metadata from Project State (elastic#131205) Add note on o11y to architecture guide (elastic#131291) Upgrade AWS Java SDK to 2.31.78 (elastic#131050) Support Fields API in conditional ingest processors (elastic#121914) ESQL - KNN function uses prefilters when pushed down to Lucene (elastic#131004) Add docs for ES|QL query logs (elastic#131287) Simplify `expectedFinalRegisterValue` computation (elastic#131274) Mute org.elasticsearch.test.rest.yaml.RcsCcsCommonYamlTestSuiteIT test {p0=search/110_field_collapsing/field collapsing, inner_hits and maxConcurrentGroupRequests} elastic#131348 ...
2 parents 7e40783 + 1da1692 commit f8b7217

File tree

280 files changed

+13241
-6472
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

280 files changed

+13241
-6472
lines changed

‎benchmarks/src/main/java/org/elasticsearch/benchmark/_nightly/esql/ValuesSourceReaderBenchmark.java‎

Lines changed: 32 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -92,7 +92,8 @@ public class ValuesSourceReaderBenchmark {
9292
"double",
9393
"keyword",
9494
"stored_keyword",
95-
"3_stored_keywords" };
95+
"3_stored_keywords",
96+
"keyword_mv" };
9697

9798
private static final int BLOCK_LENGTH = 16 * 1024;
9899
private static final int INDEX_SIZE = 10 * BLOCK_LENGTH;
@@ -332,7 +333,7 @@ public FieldNamesFieldMapper.FieldNamesFieldType fieldNames() {
332333
@Param({ "in_order", "shuffled" })
333334
public String layout;
334335

335-
@Param({ "long", "keyword", "stored_keyword" })
336+
@Param({ "long", "keyword", "stored_keyword", "keyword_mv" })
336337
public String name;
337338

338339
private Directory directory;
@@ -398,6 +399,22 @@ public void benchmark() {
398399
}
399400
}
400401
}
402+
case "keyword_mv" -> {
403+
BytesRef scratch = new BytesRef();
404+
BytesRefBlock values = op.getOutput().<BytesRefBlock>getBlock(1);
405+
for (int p = 0; p < values.getPositionCount(); p++) {
406+
int count = values.getValueCount(p);
407+
if (count > 0) {
408+
int first = values.getFirstValueIndex(p);
409+
for (int i = 0; i < count; i++) {
410+
BytesRef r = values.getBytesRef(first + i, scratch);
411+
r.offset++;
412+
r.length--;
413+
sum += Integer.parseInt(r.utf8ToString());
414+
}
415+
}
416+
}
417+
}
401418
}
402419
}
403420
long expected = 0;
@@ -407,6 +424,16 @@ public void benchmark() {
407424
expected += i % 1000;
408425
}
409426
break;
427+
case "keyword_mv":
428+
for (int i = 0; i < INDEX_SIZE; i++) {
429+
int v1 = i % 1000;
430+
expected += v1;
431+
int v2 = i % 500;
432+
if (v1 != v2) {
433+
expected += v2;
434+
}
435+
}
436+
break;
410437
case "3_stored_keywords":
411438
for (int i = 0; i < INDEX_SIZE; i++) {
412439
expected += 3 * (i % 1000);
@@ -461,7 +488,9 @@ private void setupIndex() throws IOException {
461488
new StoredField("double", (double) i),
462489
new KeywordFieldMapper.KeywordField("keyword_1", new BytesRef(c + i % 1000), keywordFieldType),
463490
new KeywordFieldMapper.KeywordField("keyword_2", new BytesRef(c + i % 1000), keywordFieldType),
464-
new KeywordFieldMapper.KeywordField("keyword_3", new BytesRef(c + i % 1000), keywordFieldType)
491+
new KeywordFieldMapper.KeywordField("keyword_3", new BytesRef(c + i % 1000), keywordFieldType),
492+
new KeywordFieldMapper.KeywordField("keyword_mv", new BytesRef(c + i % 1000), keywordFieldType),
493+
new KeywordFieldMapper.KeywordField("keyword_mv", new BytesRef(c + i % 500), keywordFieldType)
465494
)
466495
);
467496
if (i % COMMIT_INTERVAL == 0) {

‎benchmarks/src/main/java/org/elasticsearch/benchmark/compute/operator/ValuesAggregatorBenchmark.java‎

Lines changed: 5 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -95,8 +95,7 @@ static void selfTest() {
9595
try {
9696
for (String groups : ValuesAggregatorBenchmark.class.getField("groups").getAnnotationsByType(Param.class)[0].value()) {
9797
for (String dataType : ValuesAggregatorBenchmark.class.getField("dataType").getAnnotationsByType(Param.class)[0].value()) {
98-
run(Integer.parseInt(groups), dataType, 10, 0);
99-
run(Integer.parseInt(groups), dataType, 10, 1);
98+
run(Integer.parseInt(groups), dataType, 10);
10099
}
101100
}
102101
} catch (NoSuchFieldException e) {
@@ -114,10 +113,7 @@ static void selfTest() {
114113
@Param({ BYTES_REF, INT, LONG })
115114
public String dataType;
116115

117-
@Param({ "0", "1" })
118-
public int numOrdinalMerges;
119-
120-
private static Operator operator(DriverContext driverContext, int groups, String dataType, int numOrdinalMerges) {
116+
private static Operator operator(DriverContext driverContext, int groups, String dataType) {
121117
if (groups == 1) {
122118
return new AggregationOperator(
123119
List.of(supplier(dataType).aggregatorFactory(AggregatorMode.SINGLE, List.of(0)).apply(driverContext)),
@@ -132,20 +128,8 @@ private static Operator operator(DriverContext driverContext, int groups, String
132128
) {
133129
@Override
134130
public Page getOutput() {
135-
mergeOrdinal();
136131
return super.getOutput();
137132
}
138-
139-
// simulate OrdinalsGroupingOperator
140-
void mergeOrdinal() {
141-
var merged = supplier(dataType).groupingAggregatorFactory(AggregatorMode.SINGLE, List.of(1)).apply(driverContext);
142-
for (int i = 0; i < numOrdinalMerges; i++) {
143-
for (int p = 0; p < groups; p++) {
144-
merged.addIntermediateRow(p, aggregators.getFirst(), p);
145-
}
146-
}
147-
aggregators.set(0, merged);
148-
}
149133
};
150134
}
151135

@@ -352,12 +336,12 @@ private static Block groupingBlock(int groups) {
352336

353337
@Benchmark
354338
public void run() {
355-
run(groups, dataType, OP_COUNT, numOrdinalMerges);
339+
run(groups, dataType, OP_COUNT);
356340
}
357341

358-
private static void run(int groups, String dataType, int opCount, int numOrdinalMerges) {
342+
private static void run(int groups, String dataType, int opCount) {
359343
DriverContext driverContext = driverContext();
360-
try (Operator operator = operator(driverContext, groups, dataType, numOrdinalMerges)) {
344+
try (Operator operator = operator(driverContext, groups, dataType)) {
361345
Page page = page(groups, dataType);
362346
for (int i = 0; i < opCount; i++) {
363347
operator.addInput(page.shallowCopy());

‎benchmarks/src/main/java/org/elasticsearch/benchmark/vector/Int4ScorerBenchmark.java‎

Lines changed: 72 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -8,12 +8,14 @@
88
*/
99
package org.elasticsearch.benchmark.vector;
1010

11+
import org.apache.lucene.index.VectorSimilarityFunction;
1112
import org.apache.lucene.store.Directory;
1213
import org.apache.lucene.store.IOContext;
1314
import org.apache.lucene.store.IndexInput;
1415
import org.apache.lucene.store.IndexOutput;
1516
import org.apache.lucene.store.MMapDirectory;
1617
import org.apache.lucene.util.VectorUtil;
18+
import org.apache.lucene.util.quantization.OptimizedScalarQuantizer;
1719
import org.elasticsearch.common.logging.LogConfigurator;
1820
import org.elasticsearch.core.IOUtils;
1921
import org.elasticsearch.simdvec.ES91Int4VectorsScorer;
@@ -52,20 +54,26 @@ public class Int4ScorerBenchmark {
5254
LogConfigurator.configureESLogging(); // native access requires logging to be initialized
5355
}
5456

55-
@Param({ "384", "702", "1024" })
57+
@Param({ "384", "782", "1024" })
5658
int dims;
5759

58-
int numVectors = 200;
59-
int numQueries = 10;
60+
int numVectors = 20 * ES91Int4VectorsScorer.BULK_SIZE;
61+
int numQueries = 5;
6062

6163
byte[] scratch;
6264
byte[][] binaryVectors;
6365
byte[][] binaryQueries;
66+
float[] scores = new float[ES91Int4VectorsScorer.BULK_SIZE];
67+
68+
float[] scratchFloats = new float[3];
6469

6570
ES91Int4VectorsScorer scorer;
6671
Directory dir;
6772
IndexInput in;
6873

74+
OptimizedScalarQuantizer.QuantizationResult queryCorrections;
75+
float centroidDp;
76+
6977
@Setup
7078
public void setup() throws IOException {
7179
binaryVectors = new byte[numVectors][dims];
@@ -77,9 +85,19 @@ public void setup() throws IOException {
7785
binaryVector[i] = (byte) ThreadLocalRandom.current().nextInt(16);
7886
}
7987
out.writeBytes(binaryVector, 0, binaryVector.length);
88+
ThreadLocalRandom.current().nextBytes(binaryVector);
89+
out.writeBytes(binaryVector, 0, 14); // corrections
8090
}
8191
}
8292

93+
queryCorrections = new OptimizedScalarQuantizer.QuantizationResult(
94+
ThreadLocalRandom.current().nextFloat(),
95+
ThreadLocalRandom.current().nextFloat(),
96+
ThreadLocalRandom.current().nextFloat(),
97+
Short.toUnsignedInt((short) ThreadLocalRandom.current().nextInt())
98+
);
99+
centroidDp = ThreadLocalRandom.current().nextFloat();
100+
83101
in = dir.openInput("vectors", IOContext.DEFAULT);
84102
binaryQueries = new byte[numVectors][dims];
85103
for (byte[] binaryVector : binaryVectors) {
@@ -105,18 +123,66 @@ public void scoreFromArray(Blackhole bh) throws IOException {
105123
in.seek(0);
106124
for (int i = 0; i < numVectors; i++) {
107125
in.readBytes(scratch, 0, dims);
108-
bh.consume(VectorUtil.int4DotProduct(binaryQueries[j], scratch));
126+
int dp = VectorUtil.int4DotProduct(binaryQueries[j], scratch);
127+
in.readFloats(scratchFloats, 0, 3);
128+
float score = scorer.applyCorrections(
129+
queryCorrections.lowerInterval(),
130+
queryCorrections.upperInterval(),
131+
queryCorrections.quantizedComponentSum(),
132+
queryCorrections.additionalCorrection(),
133+
VectorSimilarityFunction.EUCLIDEAN,
134+
centroidDp, // assuming no centroid dot product for this benchmark
135+
scratchFloats[0],
136+
scratchFloats[1],
137+
Short.toUnsignedInt(in.readShort()),
138+
scratchFloats[2],
139+
dp
140+
);
141+
bh.consume(score);
109142
}
110143
}
111144
}
112145

113146
@Benchmark
114147
@Fork(jvmArgsPrepend = { "--add-modules=jdk.incubator.vector" })
115-
public void scoreFromMemorySegmentOnlyVector(Blackhole bh) throws IOException {
148+
public void scoreFromMemorySegment(Blackhole bh) throws IOException {
116149
for (int j = 0; j < numQueries; j++) {
117150
in.seek(0);
118151
for (int i = 0; i < numVectors; i++) {
119-
bh.consume(scorer.int4DotProduct(binaryQueries[j]));
152+
bh.consume(
153+
scorer.score(
154+
binaryQueries[j],
155+
queryCorrections.lowerInterval(),
156+
queryCorrections.upperInterval(),
157+
queryCorrections.quantizedComponentSum(),
158+
queryCorrections.additionalCorrection(),
159+
VectorSimilarityFunction.EUCLIDEAN,
160+
centroidDp
161+
)
162+
);
163+
}
164+
}
165+
}
166+
167+
@Benchmark
168+
@Fork(jvmArgsPrepend = { "--add-modules=jdk.incubator.vector" })
169+
public void scoreFromMemorySegmentBulk(Blackhole bh) throws IOException {
170+
for (int j = 0; j < numQueries; j++) {
171+
in.seek(0);
172+
for (int i = 0; i < numVectors; i += ES91Int4VectorsScorer.BULK_SIZE) {
173+
scorer.scoreBulk(
174+
binaryQueries[j],
175+
queryCorrections.lowerInterval(),
176+
queryCorrections.upperInterval(),
177+
queryCorrections.quantizedComponentSum(),
178+
queryCorrections.additionalCorrection(),
179+
VectorSimilarityFunction.EUCLIDEAN,
180+
centroidDp,
181+
scores
182+
);
183+
for (float score : scores) {
184+
bh.consume(score);
185+
}
120186
}
121187
}
122188
}

‎build-tools-internal/src/main/java/org/elasticsearch/gradle/internal/dependencies/patches/awsv2sdk/Awsv2ClassPatcher.java‎

Lines changed: 0 additions & 61 deletions
This file was deleted.

0 commit comments

Comments
 (0)