Главная
Study mode:
on
1
Intro
2
Why talk about hash tables?
3
Performance of standard containers
4
Vs linear search
5
Why are hash tables slow?
6
Talk Preview
7
What should we measure?
8
Optimization 1/7: Integer Modulo
9
Adding ska unordered_map
10
Cache Miss Lookup
11
Optimization 2/7: Cache Misses
12
A simple unordered_map
13
Open Addressing
14
dense_hash_map performance
15
Optimization 3/7: Back to Linked Lists
16
ptr_hash_map performance
17
Optimization 4/7: Robin Hood Hashing
18
Robin Hood Hashing Performance
19
A different perspective
20
The memory overhead of ska::flat_hash_map
21
Effect of max_load_factor
22
Optimization 5/7: Google's New flat_hash_map
23
How do we make it faster?
24
How many bits does Robin Hood Hashing need?
25
Optimization 6/7: Robin Hood Hashing +SIMD
26
Robin Hood SIMD Performance
27
Chaining hash table lookups
28
Reminder: What ptr_hash_map looks like
Description:
Explore advanced hash table implementations and performance optimizations beyond std::unordered_map in this comprehensive conference talk from C++Now 2018. Dive into topics such as linear probing with Robin Hood Hashing, Google's SIMD-based optimization techniques, and node-based container improvements. Learn about recent micro-optimizations that enhance lookup times and compare the performance of various hash table implementations. Gain insights into when to choose specific hash table variants and understand the trade-offs between speed and memory usage. Discover how innovative approaches continue to push the boundaries of hash table performance, even in a field with decades of optimization history.

You Can Do Better than std - unordered_map - New Improvements to Hash Table Performance

CppNow
Add to list
0:00 / 0:00