diff --git a/java/server/src/main/java/org/softwareheritage/graph/benchmark/AccessEdge.java b/java/server/src/main/java/org/softwareheritage/graph/benchmark/AccessEdge.java
--- a/java/server/src/main/java/org/softwareheritage/graph/benchmark/AccessEdge.java
+++ b/java/server/src/main/java/org/softwareheritage/graph/benchmark/AccessEdge.java
@@ -7,7 +7,7 @@
import it.unimi.dsi.big.webgraph.LazyLongIterator;
import org.softwareheritage.graph.Graph;
-import org.softwareheritage.graph.benchmark.Common;
+import org.softwareheritage.graph.benchmark.Benchmark;
import org.softwareheritage.graph.benchmark.utils.Statistics;
import org.softwareheritage.graph.benchmark.utils.Timing;
@@ -26,11 +26,12 @@
* @param args command line arguments
*/
public static void main(String[] args) throws IOException, JSAPException {
- Common.BenchArgs benchArgs = Common.parseCommandLineArgs(args);
+ Benchmark bench = new Benchmark();
+ bench.parseCommandLineArgs(args);
- Graph graph = new Graph(benchArgs.graphPath);
+ Graph graph = new Graph(bench.args.graphPath);
- long[] nodeIds = benchArgs.random.generateNodeIds(graph, benchArgs.nbNodes);
+ long[] nodeIds = bench.args.random.generateNodeIds(graph, bench.args.nbNodes);
ArrayList<Double> timings = new ArrayList<>();
for (long nodeId : nodeIds) {
@@ -41,7 +42,7 @@
timings.add(duration);
}
- System.out.println("Used " + benchArgs.nbNodes + " random edges (results are in seconds):");
+ System.out.println("Used " + bench.args.nbNodes + " random edges (results are in seconds):");
Statistics stats = new Statistics(timings);
stats.printAll();
}
diff --git a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Common.java b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Benchmark.java
rename from java/server/src/main/java/org/softwareheritage/graph/benchmark/Common.java
rename to java/server/src/main/java/org/softwareheritage/graph/benchmark/Benchmark.java
--- a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Common.java
+++ b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Benchmark.java
@@ -1,5 +1,9 @@
package org.softwareheritage.graph.benchmark;
+import java.io.BufferedWriter;
+import java.io.FileWriter;
+import java.io.IOException;
+import java.io.Writer;
import java.util.ArrayList;
import java.util.function.Function;
@@ -25,33 +29,46 @@
* @since 0.0.1
*/
-public class Common {
+public class Benchmark {
/**
- * Benchmark input arguments.
+ * Input arguments.
*/
- public static class BenchArgs {
+ public class Args {
/** Basename of the compressed graph */
public String graphPath;
/** Number of random nodes to use for the benchmark */
public int nbNodes;
+ /** File name for CSV format benchmark log */
+ public String logFile;
/** Random generator */
public Random random;
}
+ /** Command line arguments */
+ public Args args;
+
+ /**
+ * Constructor.
+ */
+ public Benchmark() {
+ this.args = new Args();
+ }
+
/**
* Parses benchmark command line arguments.
*
* @param args command line arguments
- * @return parsed arguments as a {@link BenchArgs}
*/
- public static BenchArgs parseCommandLineArgs(String[] args) throws JSAPException {
- SimpleJSAP jsap = new SimpleJSAP(Common.class.getName(),
+ public void parseCommandLineArgs(String[] args) throws JSAPException {
+ SimpleJSAP jsap = new SimpleJSAP(Benchmark.class.getName(),
"Benchmark tool for Software Heritage use-cases scenarios.",
new Parameter[] {
new UnflaggedOption("graphPath", JSAP.STRING_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED,
JSAP.NOT_GREEDY, "The basename of the compressed graph."),
new FlaggedOption("nbNodes", JSAP.INTEGER_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, 'n',
"nb-nodes", "Number of random nodes used to do the benchmark."),
+ new FlaggedOption("logFile", JSAP.STRING_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, 'l',
+ "log-file", "File name to output CSV format benchmark log."),
new FlaggedOption("seed", JSAP.LONG_PARSER, JSAP.NO_DEFAULT, JSAP.NOT_REQUIRED, 's',
"seed", "Random generator seed."),
});
@@ -61,17 +78,16 @@
System.exit(1);
}
- BenchArgs benchArgs = new BenchArgs();
- benchArgs.graphPath = config.getString("graphPath");
- benchArgs.nbNodes = config.getInt("nbNodes");
- benchArgs.random = config.contains("seed") ? new Random(config.getLong("seed")) : new Random();
-
- return benchArgs;
+ this.args.graphPath = config.getString("graphPath");
+ this.args.nbNodes = config.getInt("nbNodes");
+ this.args.logFile = config.getString("logFile");
+ this.args.random = config.contains("seed") ? new Random(config.getLong("seed")) : new Random();
}
/**
- * Times a specific endpoint and prints aggregated statistics.
+ * Times a specific endpoint and outputs individual datapoints along with aggregated statistics.
*
+ * @param useCaseName benchmark use-case name
* @param graph compressed graph used in the benchmark
* @param nodeIds node ids to use as starting point for the endpoint traversal
* @param operation endpoint function to benchmark
@@ -79,24 +95,33 @@
* href="https://docs.softwareheritage.org/devel/swh-graph/api.html#walk">API</a>
* @param algorithm traversal algorithm used in endpoint call (either "dfs" or "bfs")
*/
- public static void timeEndpoint(Graph graph, long[] nodeIds,
- Function<Endpoint.Input, Endpoint.Output> operation, String dstFmt, String algorithm) {
+ public void timeEndpoint(String useCaseName, Graph graph, long[] nodeIds,
+ Function<Endpoint.Input, Endpoint.Output> operation, String dstFmt, String algorithm)
+ throws IOException {
ArrayList<Double> timings = new ArrayList<>();
ArrayList<Double> timingsNormalized = new ArrayList<>();
- for (long nodeId : nodeIds) {
- SwhPID swhPID = graph.getSwhPID(nodeId);
+ try (Writer csvLog = new BufferedWriter(new FileWriter(args.logFile))) {
+ for (long nodeId : nodeIds) {
+ SwhPID swhPID = graph.getSwhPID(nodeId);
- Endpoint.Output output = (dstFmt == null)
- ? operation.apply(new Endpoint.Input(swhPID))
- : operation.apply(new Endpoint.Input(swhPID, dstFmt, algorithm));
+ Endpoint.Output output = (dstFmt == null)
+ ? operation.apply(new Endpoint.Input(swhPID))
+ : operation.apply(new Endpoint.Input(swhPID, dstFmt, algorithm));
- timings.add(output.meta.timings.traversal);
- if (output.meta.nbEdgesAccessed != 0) {
- timingsNormalized.add(output.meta.timings.traversal / output.meta.nbEdgesAccessed);
+ csvLog.write(String.format("%s,%s,%d,%f,%f,%f\n", useCaseName, swhPID,
+ output.meta.nbEdgesAccessed, output.meta.timings.traversal,
+ output.meta.timings.pid2node, output.meta.timings.node2pid));
+
+ timings.add(output.meta.timings.traversal);
+ if (output.meta.nbEdgesAccessed != 0) {
+ timingsNormalized.add(output.meta.timings.traversal / output.meta.nbEdgesAccessed);
+ }
}
}
+ System.out.println("\n" + useCaseName + " use-case:");
+
System.out.println("timings:");
Statistics stats = new Statistics(timings);
stats.printAll();
@@ -109,8 +134,8 @@
/**
* Same as {@link timeEndpoint} but without destination or algorithm specified to endpoint call.
*/
- public static void timeEndpoint(
- Graph graph, long[] nodeIds, Function<Endpoint.Input, Endpoint.Output> operation) {
- timeEndpoint(graph, nodeIds, operation, null, null);
+ public void timeEndpoint(String useCaseName, Graph graph, long[] nodeIds,
+ Function<Endpoint.Input, Endpoint.Output> operation) throws IOException {
+ timeEndpoint(useCaseName, graph, nodeIds, operation, null, null);
}
}
diff --git a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Browsing.java b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Browsing.java
--- a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Browsing.java
+++ b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Browsing.java
@@ -7,7 +7,7 @@
import org.softwareheritage.graph.Endpoint;
import org.softwareheritage.graph.Graph;
import org.softwareheritage.graph.Node;
-import org.softwareheritage.graph.benchmark.Common;
+import org.softwareheritage.graph.benchmark.Benchmark;
/**
* Benchmark Software Heritage <a
@@ -26,24 +26,22 @@
* @param args command line arguments
*/
public static void main(String[] args) throws IOException, JSAPException {
- Common.BenchArgs benchArgs = Common.parseCommandLineArgs(args);
+ Benchmark bench = new Benchmark();
+ bench.parseCommandLineArgs(args);
- Graph graph = new Graph(benchArgs.graphPath);
+ Graph graph = new Graph(bench.args.graphPath);
long[] dirNodeIds =
- benchArgs.random.generateNodeIdsOfType(graph, benchArgs.nbNodes, Node.Type.DIR);
+ bench.args.random.generateNodeIdsOfType(graph, bench.args.nbNodes, Node.Type.DIR);
long[] revNodeIds =
- benchArgs.random.generateNodeIdsOfType(graph, benchArgs.nbNodes, Node.Type.REV);
+ bench.args.random.generateNodeIdsOfType(graph, bench.args.nbNodes, Node.Type.REV);
Endpoint dirEndpoint = new Endpoint(graph, "forward", "dir:cnt,dir:dir");
Endpoint revEndpoint = new Endpoint(graph, "forward", "rev:rev");
- System.out.println("Used " + benchArgs.nbNodes + " random nodes (results are in seconds):");
- System.out.println("\n'ls' use-case");
- Common.timeEndpoint(graph, dirNodeIds, dirEndpoint::neighbors);
- System.out.println("\n'ls -R' use-case");
- Common.timeEndpoint(graph, dirNodeIds, dirEndpoint::visitPaths);
- System.out.println("\n'git log' use-case");
- Common.timeEndpoint(graph, revNodeIds, revEndpoint::visitNodes);
+ System.out.println("Used " + bench.args.nbNodes + " random nodes (results are in seconds):");
+ bench.timeEndpoint("ls", graph, dirNodeIds, dirEndpoint::neighbors);
+ bench.timeEndpoint("ls -R", graph, dirNodeIds, dirEndpoint::visitPaths);
+ bench.timeEndpoint("git log", graph, revNodeIds, revEndpoint::visitNodes);
}
}
diff --git a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Provenance.java b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Provenance.java
--- a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Provenance.java
+++ b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Provenance.java
@@ -6,7 +6,7 @@
import org.softwareheritage.graph.Endpoint;
import org.softwareheritage.graph.Graph;
-import org.softwareheritage.graph.benchmark.Common;
+import org.softwareheritage.graph.benchmark.Benchmark;
/**
* Benchmark Software Heritage <a
@@ -25,29 +25,30 @@
* @param args command line arguments
*/
public static void main(String[] args) throws IOException, JSAPException {
- Common.BenchArgs benchArgs = Common.parseCommandLineArgs(args);
+ Benchmark bench = new Benchmark();
+ bench.parseCommandLineArgs(args);
- Graph graph = new Graph(benchArgs.graphPath);
+ Graph graph = new Graph(bench.args.graphPath);
- long[] nodeIds = benchArgs.random.generateNodeIds(graph, benchArgs.nbNodes);
+ long[] nodeIds = bench.args.random.generateNodeIds(graph, bench.args.nbNodes);
Endpoint commitProvenanceEndpoint = new Endpoint(graph, "backward", "dir:dir,cnt:dir,dir:rev");
Endpoint originProvenanceEndpoint = new Endpoint(graph, "backward", "*");
- System.out.println("Used " + benchArgs.nbNodes + " random nodes (results are in seconds):");
-
- System.out.println("\n'commit provenance' use-case (using dfs)");
- Common.timeEndpoint(graph, nodeIds, commitProvenanceEndpoint::walk, "rev", "dfs");
- System.out.println("\n'commit provenance' use-case (using bfs)");
- Common.timeEndpoint(graph, nodeIds, commitProvenanceEndpoint::walk, "rev", "bfs");
- System.out.println("\n'complete commit provenance' use-case");
- Common.timeEndpoint(graph, nodeIds, commitProvenanceEndpoint::leaves);
-
- System.out.println("\n'origin provenance' use-case (using dfs)");
- Common.timeEndpoint(graph, nodeIds, originProvenanceEndpoint::walk, "ori", "dfs");
- System.out.println("\n'origin provenance' use-case (using bfs)");
- Common.timeEndpoint(graph, nodeIds, originProvenanceEndpoint::walk, "ori", "bfs");
- System.out.println("\n'complete origin provenance' use-case");
- Common.timeEndpoint(graph, nodeIds, originProvenanceEndpoint::leaves);
+ System.out.println("Used " + bench.args.nbNodes + " random nodes (results are in seconds):");
+
+ bench.timeEndpoint(
+ "commit provenance (dfs)", graph, nodeIds, commitProvenanceEndpoint::walk, "rev", "dfs");
+ bench.timeEndpoint(
+ "commit provenance (bfs)", graph, nodeIds, commitProvenanceEndpoint::walk, "rev", "bfs");
+ bench.timeEndpoint(
+ "complete commit provenance", graph, nodeIds, commitProvenanceEndpoint::leaves);
+
+ bench.timeEndpoint(
+ "origin provenance (dfs)", graph, nodeIds, originProvenanceEndpoint::walk, "ori", "dfs");
+ bench.timeEndpoint(
+ "origin provenance (bfs)", graph, nodeIds, originProvenanceEndpoint::walk, "ori", "bfs");
+ bench.timeEndpoint(
+ "complete origin provenance", graph, nodeIds, originProvenanceEndpoint::leaves);
}
}
diff --git a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Vault.java b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Vault.java
--- a/java/server/src/main/java/org/softwareheritage/graph/benchmark/Vault.java
+++ b/java/server/src/main/java/org/softwareheritage/graph/benchmark/Vault.java
@@ -6,7 +6,7 @@
import org.softwareheritage.graph.Endpoint;
import org.softwareheritage.graph.Graph;
-import org.softwareheritage.graph.benchmark.Common;
+import org.softwareheritage.graph.benchmark.Benchmark;
/**
* Benchmark Software Heritage <a
@@ -25,16 +25,16 @@
* @param args command line arguments
*/
public static void main(String[] args) throws IOException, JSAPException {
- Common.BenchArgs benchArgs = Common.parseCommandLineArgs(args);
+ Benchmark bench = new Benchmark();
+ bench.parseCommandLineArgs(args);
- Graph graph = new Graph(benchArgs.graphPath);
+ Graph graph = new Graph(bench.args.graphPath);
- long[] nodeIds = benchArgs.random.generateNodeIds(graph, benchArgs.nbNodes);
+ long[] nodeIds = bench.args.random.generateNodeIds(graph, bench.args.nbNodes);
Endpoint endpoint = new Endpoint(graph, "forward", "*");
- System.out.println("Used " + benchArgs.nbNodes + " random nodes (results are in seconds):");
- System.out.println("\n'git bundle' use-case");
- Common.timeEndpoint(graph, nodeIds, endpoint::visitNodes);
+ System.out.println("Used " + bench.args.nbNodes + " random nodes (results are in seconds):");
+ bench.timeEndpoint("git bundle", graph, nodeIds, endpoint::visitNodes);
}
}