From 7ecb867c4cd6916b6cb12f2ece1a4c88591ad5b5 Mon Sep 17 00:00:00 2001
From: Xiangrui Meng <meng@databricks.com>
Date: Wed, 13 Aug 2014 16:20:49 -0700
Subject: [PATCH] [MLLIB] use Iterator.fill instead of Array.fill
Iterator.fill uses less memory
Author: Xiangrui Meng <meng@databricks.com>
Closes #1930 from mengxr/rand-gen-iter and squashes the following commits:
24178ca [Xiangrui Meng] use Iterator.fill instead of Array.fill
---
.../scala/org/apache/spark/mllib/rdd/RandomRDD.scala | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/mllib/src/main/scala/org/apache/spark/mllib/rdd/RandomRDD.scala b/mllib/src/main/scala/org/apache/spark/mllib/rdd/RandomRDD.scala
index c8db3910c6..910eff9540 100644
--- a/mllib/src/main/scala/org/apache/spark/mllib/rdd/RandomRDD.scala
+++ b/mllib/src/main/scala/org/apache/spark/mllib/rdd/RandomRDD.scala
@@ -105,16 +105,16 @@ private[mllib] object RandomRDD {
def getPointIterator[T: ClassTag](partition: RandomRDDPartition[T]): Iterator[T] = {
val generator = partition.generator.copy()
generator.setSeed(partition.seed)
- Array.fill(partition.size)(generator.nextValue()).toIterator
+ Iterator.fill(partition.size)(generator.nextValue())
}
// The RNG has to be reset every time the iterator is requested to guarantee same data
// every time the content of the RDD is examined.
- def getVectorIterator(partition: RandomRDDPartition[Double],
- vectorSize: Int): Iterator[Vector] = {
+ def getVectorIterator(
+ partition: RandomRDDPartition[Double],
+ vectorSize: Int): Iterator[Vector] = {
val generator = partition.generator.copy()
generator.setSeed(partition.seed)
- Array.fill(partition.size)(new DenseVector(
- (0 until vectorSize).map { _ => generator.nextValue() }.toArray)).toIterator
+ Iterator.fill(partition.size)(new DenseVector(Array.fill(vectorSize)(generator.nextValue())))
}
}
--
GitLab