add config options for regression testing (#609)

* add config options for regression testing

* fixed map error

* make index storage directory configurable

* Revert "make index storage directory configurable"

This reverts commit 08779be.

* move disk usage to summary and remove from table

Author: MarkWolters

jvector-examples/src/main/java/io/github/jbellis/jvector/example/Grid.java

@@ -242,6 +242,9 @@ static void runOneGraph(OnDiskGraphIndexCache cache,
         // Capture post-build memory and disk state
         diagnostics.capturePostPhaseSnapshot("Graph Build");
 
+        diagnostics.printDiskStatistics("Graph Index Build");
+        System.out.printf("Index build time: %f seconds%n", Grid.getIndexBuildTimeSeconds(ds.getName()));
+
         try {
             for (var cpSupplier : compressionGrid) {
                 indexes.forEach((features, index) -> {
@@ -681,10 +684,9 @@ public static List<BenchResult> runAllAndCollectResults(
                                         Path testDirectory = Files.createTempDirectory("bench");
                                         try {
                                             // Capture initial state
-                                            var diagnostics = new io.github.jbellis.jvector.example.benchmarks.diagnostics.BenchmarkDiagnostics(getDiagnosticLevel());
+                                            var diagnostics = new BenchmarkDiagnostics(getDiagnosticLevel());
                                             diagnostics.setMonitoredDirectory(testDirectory);
                                             diagnostics.capturePrePhaseSnapshot("Build");
-
                                             Map<Set<FeatureId>, ImmutableGraphIndex> indexes = new HashMap<>();
 
                                             var compressor = getCompressor(buildCompressor, ds);
@@ -732,6 +734,7 @@ public static List<BenchResult> runAllAndCollectResults(
 
                                             // Capture post-build state
                                             diagnostics.capturePostPhaseSnapshot("Build");
+                                            diagnostics.printDiskStatistics("Graph Index Build");
                                             var buildSnapshot = diagnostics.getLatestSystemSnapshot();
                                             var buildDiskSnapshot = diagnostics.getLatestDiskSnapshot();
 

jvector-examples/src/main/java/io/github/jbellis/jvector/example/benchmarks/QueryTester.java

@@ -112,22 +112,6 @@ public List<Metric> run(
             results.add(Metric.of("Max offheap usage", ".1f",
                 systemSnapshot.memoryStats.getTotalOffHeapMemory() / (1024.0 * 1024.0)));
         }
-        
-        if (diskSnapshot != null) {
-            // Total file size in MB
-            results.add(Metric.of("Total file size", ".1f",
-                diskSnapshot.totalBytes / (1024.0 * 1024.0)));
-            
-            // Number of files
-            results.add(Metric.of("Number of files", ".0f",
-                (double) diskSnapshot.fileCount));
-        }
-        
-        // Add index build time if available
-        if (datasetName != null && Grid.getIndexBuildTimeSeconds(datasetName) != null) {
-            results.add(Metric.of("Index build time", ".2f",
-                Grid.getIndexBuildTimeSeconds(datasetName)));
-        }
 
         return results;
     }

jvector-examples/src/main/java/io/github/jbellis/jvector/example/benchmarks/diagnostics/BenchmarkDiagnostics.java

@@ -259,7 +259,6 @@ public void logSummary() {
         System.out.println("\n=== BENCHMARK DIAGNOSTICS SUMMARY ===");
         System.out.printf("Diagnostic Level: %s%n", level);
         System.out.printf("System Snapshots: %d%n", snapshots.size());
-        System.out.printf("Disk Snapshots: %d%n", diskSnapshots.size());
         System.out.printf("Timing Analyses: %d%n", timingAnalyses.size());
 
         if (!snapshots.isEmpty()) {
@@ -285,28 +284,31 @@ public void logSummary() {
                 DiskUsageMonitor.formatBytes(last.memoryStats.mappedBufferMemory));
         }
 
+        // Performance variance analysis
+        if (timingAnalyses.size() > 1) {
+            System.out.println("\nPerformance Variance Analysis:");
+            for (int i = 1; i < timingAnalyses.size(); i++) {
+                comparePhases(timingAnalyses.get(0).phase, timingAnalyses.get(i).phase);
+            }
+        }
+
+        System.out.println("=====================================\n");
+    }
+
+    public void printDiskStatistics(String label) {
         // Disk usage summary
         if (!diskSnapshots.isEmpty()) {
             DiskUsageMonitor.DiskUsageSnapshot firstDisk = diskSnapshots.get(0);
             DiskUsageMonitor.DiskUsageSnapshot lastDisk = diskSnapshots.get(diskSnapshots.size() - 1);
             DiskUsageMonitor.DiskUsageSnapshot totalDisk = lastDisk.subtract(firstDisk);
 
-            System.out.printf("\nDisk Usage Summary:%n");
+            System.out.printf("\nDisk Usage Summary %s:%n", label);
             System.out.printf("  Total Disk Used: %s%n", DiskUsageMonitor.formatBytes(lastDisk.totalBytes));
             System.out.printf("  Total Files: %d%n", lastDisk.fileCount);
             System.out.printf("  Net Change: %s, %+d files%n",
-                DiskUsageMonitor.formatBytes(totalDisk.totalBytes), totalDisk.fileCount);
-        }
-
-        // Performance variance analysis
-        if (timingAnalyses.size() > 1) {
-            System.out.println("\nPerformance Variance Analysis:");
-            for (int i = 1; i < timingAnalyses.size(); i++) {
-                comparePhases(timingAnalyses.get(0).phase, timingAnalyses.get(i).phase);
-            }
+                    DiskUsageMonitor.formatBytes(totalDisk.totalBytes), totalDisk.fileCount);
         }
 
-        System.out.println("=====================================\n");
     }
 
     /**