Demonstrate overheads for various Java Collection implementations. This
measures storing int data, excluding the cost of boxed Integers.
This compares built-in Java, fastutil,
Guava, and
Trove collections.
Tested with JDK 8 on x86-64 Linux with
compressed Oops
(default).
Measure per-entry overhead by creating a single large collection. Measure
per-instance overhead by creating many small collections with a single element.
This over-reports per-instance overhead by 4 bytes due to
Collection<Collection>
| Collection | Per-entry Overhead (bytes) |
Per-instance Overhead (bytes) |
|---|---|---|
| List | ||
| ArrayList | 4 | 52 |
| ImmutableList | 4 | 20 |
| LinkedList | 24 | 60 |
| TIntArrayList | 4 | 52 |
| TIntLinkedList | 24 | 60 |
| Map | ||
| Cache | 56 | 916 |
| ConcurrentHashMap | 40 | 180 |
| ConcurrentSkipListMap | 36 | 145 |
| EnumMap | n/a | 124 |
| HashMap | 40 | 108 |
| ImmutableMap | 32 | 44 |
| ImmutableRangeMap | 64 | 116 |
| ImmutableSortedMap | 8 | 108 |
| Int2IntAVLTreeMap | 32 | 164 |
| Int2IntLinkedOpenHashMap | 32 | 188 |
| Int2IntOpenHashMap | 16 | 132 |
| Int2IntRBTreeMap | 32 | 452 |
| LinkedHashMap | 48 | 124 |
| LongLongHashMap | 32 | 116 |
| MapMaker | 40 | 244 |
| TIntIntHashMap | 18 | 156 |
| TreeMap | 40 | 92 |
| TreeRangeMap | 120 | 188 |
| Multiset | ||
| ConcurrentHashMultiset | 56 | 220 |
| HashMultiset | 56 | 156 |
| ImmutableMultiset | 32 | 116 |
| LinkedHashMultiset | 64 | 172 |
| TreeMultiset | 56 | 212 |
| Queue | ||
| ArrayDeque | 8 | 108 |
| ConcurrentLinkedDeque | 24 | 76 |
| PriorityQueue | 4 | 60 |
| Set | ||
| ConcurrentSkipListSet | 36 | 161 |
| EnumSet | n/a | 36 |
| HashSet | 40 | 124 |
| ImmutableRangeSet | 60 | 100 |
| ImmutableSet | 12 | 28 |
| ImmutableSortedSet | 4 | 52 |
| IntAVLTreeSet | 32 | 140 |
| IntLinkedOpenHashSet | 24 | 132 |
| IntOpenHashSet | 8 | 84 |
| IntRBTreeSet | 32 | 436 |
| LinkedHashSet | 48 | 140 |
| LongHashSet | 16 | 76 |
| TIntHashSet | 8 | 116 |
| TreeRangeSet | 96 | 180 |
| TreeSet | 40 | 108 |
Collected via:
target/java-collection-overhead --list | grep -v Enum | while read c; do target/java-collection-overhead $c $((8 * 1024 * 1024)) 1 || break; done
target/java-collection-overhead --list | while read c; do target/java-collection-overhead $c 1 $((1024 * 1024)) || break; done- https://github.com/DimitrisAndreou/memory-measurer/blob/master/ElementCostInDataStructures.txt
- https://www.ibm.com/developerworks/java/library/j-codetoheap/
Copyright (C) 2018 Andrew Gaul
Licensed under the Apache License, Version 2.0