Skip to content

Partitioned testing #661

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

Draft
jshook wants to merge 7 commits into
base: main
Choose a base branch
from
Draft

Partitioned testing #661

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
aeac2a4
support named entries.yaml files
jshook Apr 23, 2026
d6b52c9
scaffolding for additional partitioned testing
jshook Apr 23, 2026
e189e9f
multiple grid repetitions/query runs
jshook Apr 23, 2026
2a4bea0
glob support in dataset metadata
jshook Apr 23, 2026
fbbc855
allow index parameters to be also_for others by pattern
jshook Apr 23, 2026
20db6eb
add alpha to index construction parameters
jshook Apr 23, 2026
b28cfa0
Fan-out on compression parameters
jshook Apr 23, 2026
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
34 changes: 22 additions & 12 deletions jvector-examples/src/main/java/io/github/jbellis/jvector/example/AutoBenchYAML.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -140,18 +140,28 @@ public static void main(String[] args) throws IOException {
}
logger.info("Using configuration: {}", config);

List<BenchResult> datasetResults = Grid.runAllAndCollectResults(ds,
config.construction.useSavedIndexIfExists,
config.construction.outDegree,
config.construction.efConstruction,
config.construction.neighborOverflow,
config.construction.addHierarchy,
config.construction.refineFinalGraph,
config.construction.getFeatureSets(),
config.construction.getCompressorParameters(),
config.search.getCompressorParameters(),
config.search.topKOverquery,
config.search.useSearchPruning);
List<BenchResult> datasetResults = new ArrayList<>();
int repetitions = config.repetitionsOrOne();
for (int rep = 0; rep < repetitions; rep++) {
if (repetitions > 1) {
logger.info("{}: repetition {} of {}", datasetName, rep + 1, repetitions);
}
datasetResults.addAll(Grid.runAllAndCollectResults(ds,
config.construction.useSavedIndexIfExists,
config.construction.outDegree,
config.construction.efConstruction,
config.construction.neighborOverflow,
config.construction.alpha,
config.construction.addHierarchy,
config.construction.refineFinalGraph,
config.construction.getFeatureSets(),
config.construction.getCompressorParameters(),
config.search.getCompressorParameters(),
config.search.topKOverquery,
config.search.useSearchPruning,
config.search.queryRunsOrDefault(),
rep));
}
results.addAll(datasetResults);

logger.info("Benchmark completed for dataset: {}", datasetName);
Expand Down
7 changes: 4 additions & 3 deletions jvector-examples/src/main/java/io/github/jbellis/jvector/example/Bench.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,7 @@ public static void main(String[] args) throws IOException {
List.of(1.0, 2.0) // oq
); // rerankK = oq * topK
var neighborOverflowGrid = List.of(1.2f); // List.of(1.2f, 2.0f);
var alphaGrid = List.of(1.2f); // List.of(1.0f, 1.2f, 1.4f);
var addHierarchyGrid = List.of(true); // List.of(false, true);
var refineFinalGraphGrid = List.of(true); // List.of(false, true);
var usePruningGrid = List.of(true); // List.of(false, true);
Expand Down Expand Up @@ -82,10 +83,10 @@ public static void main(String[] args) throws IOException {
// compile regex and do substring matching using find
var pattern = Pattern.compile(regex);

execute(pattern, enableIndexCache, buildCompression, featureSets, searchCompression, mGrid, efConstructionGrid, neighborOverflowGrid, addHierarchyGrid, refineFinalGraphGrid, topKGrid, usePruningGrid);
execute(pattern, enableIndexCache, buildCompression, featureSets, searchCompression, mGrid, efConstructionGrid, neighborOverflowGrid, alphaGrid, addHierarchyGrid, refineFinalGraphGrid, topKGrid, usePruningGrid);
}

private static void execute(Pattern pattern, boolean enableIndexCache, List<Function<DataSet, CompressorParameters>> buildCompression, List<EnumSet<FeatureId>> featureSets, List<Function<DataSet, CompressorParameters>> compressionGrid, List<Integer> mGrid, List<Integer> efConstructionGrid, List<Float> neighborOverflowGrid, List<Boolean> addHierarchyGrid, List<Boolean> refineFinalGraphGrid, Map<Integer, List<Double>> topKGrid, List<Boolean> usePruningGrid) throws IOException {
private static void execute(Pattern pattern, boolean enableIndexCache, List<Function<DataSet, CompressorParameters>> buildCompression, List<EnumSet<FeatureId>> featureSets, List<Function<DataSet, CompressorParameters>> compressionGrid, List<Integer> mGrid, List<Integer> efConstructionGrid, List<Float> neighborOverflowGrid, List<Float> alphaGrid, List<Boolean> addHierarchyGrid, List<Boolean> refineFinalGraphGrid, Map<Integer, List<Double>> topKGrid, List<Boolean> usePruningGrid) throws IOException {
var datasetCollection = DatasetCollection.load();
var datasetNames = datasetCollection.getAll().stream().filter(dn -> pattern.matcher(dn).find()).collect(Collectors.toList());
System.out.println("Executing the following datasets: " + datasetNames);
Expand All @@ -94,7 +95,7 @@ private static void execute(Pattern pattern, boolean enableIndexCache, List<Func
DataSet ds = DataSets.loadDataSet(datasetName).orElseThrow(
() -> new RuntimeException("Dataset " + datasetName + " not found")
).getDataSet();
Grid.runAll(ds, enableIndexCache, mGrid, efConstructionGrid, neighborOverflowGrid, addHierarchyGrid, refineFinalGraphGrid, featureSets, buildCompression, compressionGrid, topKGrid, usePruningGrid);
Grid.runAll(ds, enableIndexCache, mGrid, efConstructionGrid, neighborOverflowGrid, alphaGrid, addHierarchyGrid, refineFinalGraphGrid, featureSets, buildCompression, compressionGrid, topKGrid, usePruningGrid);
}
}
}
41 changes: 27 additions & 14 deletions jvector-examples/src/main/java/io/github/jbellis/jvector/example/BenchYAML.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@
import io.github.jbellis.jvector.example.benchmarks.datasets.DataSets;
import io.github.jbellis.jvector.example.reporting.RunArtifacts;
import io.github.jbellis.jvector.example.reporting.SearchReportingCatalog;
import io.github.jbellis.jvector.example.util.BenchArgExpander;
import io.github.jbellis.jvector.example.yaml.DatasetCollection;
import io.github.jbellis.jvector.example.yaml.MultiConfig;
import io.github.jbellis.jvector.example.yaml.RunConfig;
Expand All @@ -43,9 +44,10 @@ public static void main(String[] args) throws IOException {
if (args == null) {
throw new InvalidParameterException("argv[] is null, check your maven exec config");
}
// Expand range shorthand like sift1m:label_[00..11] into 12 separate tokens
args = BenchArgExpander.expandAll(args);
String regex = Arrays.stream(args)
.filter(Objects::nonNull)
.flatMap(s -> Arrays.stream(s.split("\\s")))
.filter(s -> !s.isEmpty())
.collect(Collectors.joining("|"));
var pattern = Pattern.compile(regex.isEmpty() ? ".*" : regex);
Expand Down Expand Up @@ -124,20 +126,31 @@ public static void main(String[] args) throws IOException {
// Register dataset info the first time we actually load the dataset for benchmarking
artifacts.registerDataset(datasetName, ds);

Grid.runAll(ds,
config.construction.useSavedIndexIfExists,
config.construction.outDegree,
config.construction.efConstruction,
config.construction.neighborOverflow,
config.construction.addHierarchy,
config.construction.refineFinalGraph,
config.construction.getFeatureSets(),
config.construction.getCompressorParameters(),
config.search.getCompressorParameters(),
config.search.topKOverquery,
config.search.useSearchPruning,
artifacts);
int repetitions = config.repetitionsOrOne();
for (int rep = 0; rep < repetitions; rep++) {
if (repetitions > 1) {
System.out.printf("%s: repetition %d of %d%n", datasetName, rep + 1, repetitions);
}
Grid.runAll(ds,
config.construction.useSavedIndexIfExists,
config.construction.outDegree,
config.construction.efConstruction,
config.construction.neighborOverflow,
config.construction.alpha,
config.construction.addHierarchy,
config.construction.refineFinalGraph,
config.construction.getFeatureSets(),
config.construction.getCompressorParameters(),
config.search.getCompressorParameters(),
config.search.topKOverquery,
config.search.useSearchPruning,
config.search.queryRunsOrDefault(),
rep,
artifacts);
}
}

artifacts.printSummary();
}

private static String wrapList(java.util.List<String> items, int perLine, String indent) {
Expand Down
Loading
Loading