It’s a drop-in replacement of OTP sets module based on maps. It’s
much faster and simpler, as it makes heavy use of maps BIFs.
sets interface and sematics are fully preserved, so it should be
sufficient to just use replace sets with map_sets in your codebase
via sed.
A few unscientific crude benchmarks:
Operating System: Linux
CPU Information: Intel(R) Core(TM) i5-4460 CPU @ 3.20GHz
Number of Available Cores: 4
Available memory: 7.74 GB
Elixir 1.7.4
Erlang 20.3.8.9
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 10 s
memory time: 2 s
parallel: 1
inputs: none specified
Estimated total run time: 28 s
Name ips average deviation median 99th %
sets from_list 215.46 4.64 ms ±17.34% 4.43 ms 7.94 ms
map_sets from_list 154.53 6.47 ms ±14.01% 6.25 ms 9.25 ms
Comparison:
sets from_list 215.46
map_sets from_list 154.53 - 1.39x slower
Memory usage statistics:
Name Memory usage
sets from_list 13.09 MB
map_sets from_list 0.46 MB - 0.04x memory usage
Benchmarking map_sets empty intersection...
Benchmarking sets empty intersection...
Name ips average deviation median 99th %
map_sets empty intersection 1.10 K 0.91 ms ±21.91% 0.84 ms 1.16 ms
sets empty intersection 0.62 K 1.62 ms ±3.55% 1.62 ms 1.74 ms
Comparison:
map_sets empty intersection 1.10 K
sets empty intersection 0.62 K - 1.79x slower
Memory usage statistics:
Name Memory usage
map_sets empty intersection 0.109 KB
sets empty intersection 173.25 KB - 1584.00x memory usage
Name ips average deviation median 99th %
sets intersection 573.98 1.74 ms ±6.34% 1.72 ms 2.15 ms
map_sets intersection 464.68 2.15 ms ±26.53% 2.15 ms 4.20 ms
Comparison:
sets intersection 573.98
map_sets intersection 464.68 - 1.24x slower
Memory usage statistics:
Name Memory usage
sets intersection 0.25 MB
map_sets intersection 1.18 MB - 4.80x memory usage
Name ips average deviation median 99th %
map_sets union 2.93 K 0.34 ms ±24.78% 0.29 ms 0.50 ms
sets union 0.23 K 4.26 ms ±10.48% 4.14 ms 5.74 ms
Comparison:
map_sets union 2.93 K
sets union 0.23 K - 12.47x slower
Memory usage statistics:
Name Memory usage
map_sets union 0.150 MB
sets union 15.00 MB - 100.30x memory usage
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent