This repository contains an implementation of uint32-to-uint32 map which is ~20-50% faster than Go standard map for the same types (see benchmarks below). The code was based of Brent Pedersen's intintmap and the main logic remains intact, with some bug fixes and improvements of the API itself. The map is backed by a single array which interleaves keys and values to improve data locality.
// Create a new map with capacity of 1024 (resizeable) and 90% desired fill rate
m := intmap.NewWithFill(1024, 0.90)
// Store a few key/value pairs
m.Store(1, 100)
m.Store(2, 200)
// Load them
v, ok := m.Load(1)
v, ok := m.Load(2)
// Delete keys
m.Delete(1)
m.Delete(2)Looking at the benchmarks agains the standard Go map, this map should perform roughly 20-50% better depending on the conditions.
cpu: 13th Gen Intel(R) Core(TM) i7-13700K
BenchmarkStore/intmap-24 144682518 8.203 ns/op 0 B/op 0 allocs/op
BenchmarkStore/sync-24 47746893 24.78 ns/op 0 B/op 0 allocs/op
BenchmarkStore/stdmap-24 65009140 17.73 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/intmap-0%-24 40558078 28.98 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/sync-0%-24 36516896 31.53 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/stdmap-0%-24 76847426 15.57 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/intmap-10%-24 40846196 28.31 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/sync-10%-24 37622625 31.21 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/stdmap-10%-24 69917145 16.90 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/intmap-50%-24 52405636 21.62 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/sync-50%-24 44567251 25.75 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/stdmap-50%-24 50478930 23.68 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/intmap-90%-24 140594277 8.486 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/sync-90%-24 83630574 14.40 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/stdmap-90%-24 69522844 17.17 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/intmap-100%-24 189147504 6.271 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/sync-100%-24 88044195 13.58 ns/op 0 B/op 0 allocs/op
BenchmarkLoad/stdmap-100%-24 78736231 15.12 ns/op 0 B/op 0 allocs/op