better initial allocation

Baunsgaard · Oct 23, 2024 · 6e9d587 · 6e9d587
1 parent 3447463
commit 6e9d587
Showing 1 changed file with 6 additions and 1 deletion.
diff --git a/src/main/java/org/apache/sysds/runtime/frame/data/columns/Array.java b/src/main/java/org/apache/sysds/runtime/frame/data/columns/Array.java
@@ -151,7 +151,8 @@ private Map<T, Integer> parallelCreateRecodeMap(int estimate, ExecutorService po
 				tasks.add(pool.submit(() -> createRecodeMap(estimate, start, end)));
 			}
 			// make the initial map thread local allocation.
-			final Map<T, Integer> map = createRecodeMap(estimate, 0, blk);
+			final Map<T, Integer> map = new HashMap<>((int)(estimate * 1.3));
+			createRecodeMap(map, 0, blk);
 			for(int i = 0; i < tasks.size(); i++) { // merge with other threads work.
 				final Map<T, Integer> map2 = tasks.get(i).get();
 				mergeRecodeMaps(map, map2);
@@ -181,6 +182,10 @@ private void mergeRecodeMaps(Map<T, Integer> target, Map<T, Integer> from) {
 	private Map<T, Integer> createRecodeMap(final int estimate, final int s, final int e) {
 		// * 1.3 because we hashMap has a load factor of 1.75
 		final Map<T, Integer> map = new HashMap<>((int) (Math.min((long) estimate, (e - s)) * 1.3));
+		return createRecodeMap(map, s, e);
+	}
+
+	private Map<T, Integer> createRecodeMap( Map<T, Integer> map, final int s, final int e) {
 		int id = 1;
 		for(int i = s; i < e; i++)
 			id = addValRecodeMap(map, id, i);