FlatMap

学习bRPC FlatMap。

NAME

FlatMap - Maybe the fastest hashmap, with tradeoff of space.

EXAMPLE

  1. #include <string>
  2. #include <butil/logging.h>
  3. #include <butil/containers/flat_map.h>
  4. void flatmap_example() {
  5. butil::FlatMap<int, std::string> map;
  6. // bucket_count: initial count of buckets, big enough to avoid resize.
  7. // load_factor: element_count * 100 / bucket_count, 80 as default.
  8. int bucket_count = 1000;
  9. int load_factor = 80;
  10. map.init(bucket_count, load_factor);
  11. map.insert(10, "hello");
  12. map[20] = "world";
  13. std::string* value = map.seek(20);
  14. CHECK(value != NULL);
  15. CHECK_EQ(2UL, map.size());
  16. CHECK_EQ(0UL, map.erase(30));
  17. CHECK_EQ(1UL, map.erase(10));
  18. LOG(INFO) << "All elements of the map:";
  19. for (butil::FlatMap<int, std::string>::const_iterator it = map.begin(); it != map.end(); ++it) {
  20. LOG(INFO) << it->first << " : " << it->second;
  21. }
  22. map.clear();
  23. CHECK_EQ(0UL, map.size());
  24. }

DESCRIPTION

FlatMap可能是最快的哈希表,但当value较大时它需要更多的内存,它最适合作为检索过程中需要极快查找的小字典。

原理:把开链桶中第一个节点的内容直接放桶内。由于在实践中,大部分桶没有冲突或冲突较少,所以大部分操作只需要一次内存跳转:通过哈希值访问对应的桶。桶内两个及以上元素仍存放在链表中,由于桶之间彼此独立,一个桶的冲突不会影响其他桶,性能很稳定。在很多时候,FlatMap的查找性能和原生数组接近。

BENCHMARK

下面是FlatMap和其他key/value容器的比较:

  • AlignHashMap:闭链中较快的实现。
  • CowHashMap:smalltable中的开链哈希表,和普通开链不同的是带Copy-on-write逻辑。
  • std::map:非哈希表,一般是红黑树。
  1. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 8 bytes ]
  2. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 15/19/30/102ns
  3. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 7/11/33/146ns
  4. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 10/28/26/93ns
  5. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 6/9/29/100ns
  6. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 10/21/26/130ns
  7. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 5/10/30/104ns
  8. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 32 bytes ]
  9. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 23/31/31/130ns
  10. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 9/11/72/104ns
  11. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 20/53/28/112ns
  12. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 7/10/29/101ns
  13. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 20/46/28/137ns
  14. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 7/10/29/112ns
  15. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 128 bytes ]
  16. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 34/109/91/179ns
  17. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 8/11/33/112ns
  18. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 28/76/86/169ns
  19. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 8/9/30/110ns
  20. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Sequentially inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 28/68/87/201ns
  21. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Sequentially erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 9/9/30/125ns
  22. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 8 bytes ]
  23. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 14/56/29/157ns
  24. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 9/11/31/181ns
  25. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 11/17/27/156ns
  26. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 6/10/30/204ns
  27. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 13/26/27/212ns
  28. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 7/11/38/309ns
  29. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 32 bytes ]
  30. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 24/32/32/181ns
  31. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 10/12/32/182ns
  32. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 21/46/35/168ns
  33. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 7/10/36/209ns
  34. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 24/46/31/240ns
  35. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 8/11/40/314ns
  36. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:474] [ value = 128 bytes ]
  37. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 100 into FlatMap/AlignHashMap/CowHashMap/std::map takes 36/114/93/231ns
  38. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 100 from FlatMap/AlighHashMap/CowHashMap/std::map takes 9/12/35/190ns
  39. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 1000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 44/94/88/224ns
  40. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 1000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 8/10/34/236ns
  41. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:521] Randomly inserting 10000 into FlatMap/AlignHashMap/CowHashMap/std::map takes 46/92/93/314ns
  42. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:558] Randomly erasing 10000 from FlatMap/AlighHashMap/CowHashMap/std::map takes 12/11/42/362ns
  43. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 8 bytes ]
  44. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/7/12/54ns
  45. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 3/7/11/78ns
  46. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/8/13/172ns
  47. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 32 bytes ]
  48. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 5/8/12/55ns
  49. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/8/11/82ns
  50. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 6/10/14/164ns
  51. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 128 bytes ]
  52. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 7/9/13/56ns
  53. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 6/10/12/93ns
  54. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 9/12/21/166ns
  55. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 8 bytes ]
  56. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/7/11/56ns
  57. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 3/7/11/79ns
  58. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/9/13/173ns
  59. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 32 bytes ]
  60. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 5/8/12/54ns
  61. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 4/8/11/100ns
  62. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 6/10/14/165ns
  63. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:576] [ value = 128 bytes ]
  64. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 100 from FlatMap/AlignHashMap/CowHashMap/std::map takes 7/9/12/56ns
  65. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 1000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 6/10/12/88ns
  66. TRACE: 12-30 13:19:53: * 0 [test_flat_map.cpp:637] Seeking 10000 from FlatMap/AlignHashMap/CowHashMap/std::map takes 9/14/20/169ns

Overview of hashmaps

哈希表是最常用的数据结构,它的基本原理是通过计算哈希值把不同的key分散到不同的区间,在查找时通过key的哈希值能快速地缩小查找区间。在使用恰当参数的前提下,哈希表在大部分时候能在O(1)时间内把一个key映射为value。像其他算法一样,这个“O(1)”在不同的实现中差异很大。哈希表的实现一般有两部分:

计算哈希值(非加密型)

即把key散列开的方法,最常见的莫过于线性同余,但一个好的哈希算法(非加密型)要考虑很多因素:

  • 结果是确定的。
  • 雪崩效应:输入中一个bit的变化应该尽量影响输出所有bit的变化。
  • 输出应尽量在值域中均匀分布。
  • 充分利用现代cpu特性:成块计算,减少分支,循环展开等等。

大部分哈希算法针对的只是一个key,不会耗用太多的cpu。影响主要来自哈希表的整体数据分布,对于工程师来说,选用何种算法得看实践效果,一些最简单的方法也许就有很好的效果。通用算法可选择Murmurhash。

解决冲突

哈希值可能重合,解决冲突是哈希表性能的另一关键因素。常见的冲突解决方法有:

  • 开链哈希(open hashing, closed addressing): 开链哈希表是链表的数组,其中链表一般称为桶。当若干个key落到同一个桶时,做链表插入。这是最通用的结构,有很多优点:占用内存为O(NumElement * (KeySize + ValueSize + SomePointers)),resize时候不会使之前的存放key/value的内存失效。桶之间是独立的,一个桶的冲突不会影响到其他桶,平均查找时间较为稳定,独立的桶也易于高并发。缺点是至少要两次内存跳转:先跳到桶入口,再跳到桶中的第一个节点。对于一些很小的表这个问题不明显,因为当表很小时,节点内存是接近的,但当表变大时,访存就愈发随机。如果一次访存在50ns左右(2G左右主频),开链哈希的查找时间往往就在100ns以上。在检索端的层层ranking过程中,对一些热点字典的查找1秒内可能有几百万次以上,开链哈希有时会成为热点。一些产品线可能对开链哈希的内存也有诟病,因为每对key/value都需要额外的指针。
  • 闭链哈希(closed hashing or open addressing): 闭链的初衷是减少内存跳转,桶不再是链表入口,而只需要记录一对key/value与一些标记,当桶被占时,按照不同的探查方法直到找到空桶为止。比如线性探查就是查找下一个桶,二次探查是按1,2,4,9…平方数位移查找。优点是:当表很空时或冲突较少时,查找只需要一次访存,也不需要管理节点内存池。但仅此而已,这个方法带来了更多缺点:桶个数必须大于元素个数,resize后之前的内存全部失效,难以并发. 更关键的是聚集效应:当区域内元素较多时(超过70%,其实不算多),大量元素的实际桶和它们应在的桶有较大位移。这使哈希表的主要操作都要扫过一大片内存才能找到元素,性能不稳定难以预测。闭链哈希表在很多人的印象中“很快”,但在复杂的应用中往往不如开链哈希表,并且可能是数量级的慢。闭链有一些衍生版本试图解决这个问题,比如Hopscotch hashing
  • 混合开链和闭链:一般是把桶数组中的一部分拿出来作为容纳冲突元素的空间,典型如Coalesced hashing,但这种结构没有解决开链的内存跳转问题,结构又比闭链复杂很多,工程效果并不好。
  • 多次哈希:一般用多个哈希表代替一个哈希表,当发生冲突时(用另一个哈希值)尝试另一个哈希表。典型如Cuckoo hashing,这个结构也没有解决内存跳转。

修改于 2024年10月7日: Oncall report (1b7065e)