concurrent-map
concurrent-map copied to clipboard
orcaman
[#136] Add custom shard count value in concurrent map
#136
Hello, what I mentioned in the above issue page, I suggest to use custom shard count for each concurrent map instance. I would ask you please review this change and let me know how do you think about it? 😃
Thanks a lot!
(Please let me know if I should post more information before posting PR. :-) )
Benchmark result
> go test -bench=. -benchtime=10s -benchmem
goos: darwin
goarch: amd64
pkg: github.com/orcaman/concurrent-map/v2
cpu: Intel(R) Core(TM) i5-10600 CPU @ 3.30GHz
BenchmarkItems-12 5437 2329913 ns/op 1938390 B/op 417 allocs/op
BenchmarkItemsInteger-12 5905 1957807 ns/op 1490132 B/op 485 allocs/op
BenchmarkItemsInt-12 6870 1766993 ns/op 1336459 B/op 460 allocs/op
BenchmarkMarshalJson-12 2079 5853171 ns/op 3125235 B/op 20438 allocs/op
BenchmarkStrconv-12 458381920 30.00 ns/op 14 B/op 1 allocs/op
BenchmarkSingleInsertAbsent-12 33966343 535.6 ns/op 161 B/op 1 allocs/op
BenchmarkSingleInsertAbsentSyncMap-12 17470662 927.0 ns/op 177 B/op 5 allocs/op
BenchmarkSingleInsertPresent-12 237099622 49.44 ns/op 0 B/op 0 allocs/op
BenchmarkSingleInsertPresentSyncMap-12 164643420 73.08 ns/op 16 B/op 1 allocs/op
BenchmarkMultiInsertDifferentSyncMap-12 2763302 4833 ns/op 946 B/op 35 allocs/op
BenchmarkMultiInsertDifferent_1_Shard-12 4301349 3238 ns/op 317 B/op 2 allocs/op
BenchmarkMultiInsertDifferent_16_Shard-12 26476600 589.3 ns/op 144 B/op 2 allocs/op
BenchmarkMultiInsertDifferent_32_Shard-12 35410960 437.2 ns/op 186 B/op 2 allocs/op
BenchmarkMultiInsertDifferent_256_Shard-12 18891256 720.2 ns/op 217 B/op 4 allocs/op
BenchmarkMultiInsertSame-12 4410878 2690 ns/op 16 B/op 1 allocs/op
BenchmarkMultiInsertSameSyncMap-12 3264914 4324 ns/op 552 B/op 31 allocs/op
BenchmarkMultiGetSame-12 23550031 495.6 ns/op 16 B/op 1 allocs/op
BenchmarkMultiGetSameSyncMap-12 52082953 238.2 ns/op 16 B/op 1 allocs/op
BenchmarkMultiGetSetDifferentSyncMap-12 2035042 6852 ns/op 959 B/op 37 allocs/op
BenchmarkMultiGetSetDifferent_1_Shard-12 3950667 4776 ns/op 437 B/op 5 allocs/op
BenchmarkMultiGetSetDifferent_16_Shard-12 5013445 2566 ns/op 221 B/op 4 allocs/op
BenchmarkMultiGetSetDifferent_32_Shard-12 15109965 2375 ns/op 255 B/op 4 allocs/op
BenchmarkMultiGetSetDifferent_256_Shard-12 18408936 722.6 ns/op 221 B/op 4 allocs/op
BenchmarkMultiGetSetBlockSyncMap-12 20848772 689.7 ns/op 544 B/op 32 allocs/op
BenchmarkMultiGetSetBlock_1_Shard-12 4526481 6174 ns/op 217 B/op 3 allocs/op
BenchmarkMultiGetSetBlock_16_Shard-12 6353364 2003 ns/op 68 B/op 2 allocs/op
BenchmarkMultiGetSetBlock_32_Shard-12 11579113 1180 ns/op 66 B/op 2 allocs/op
BenchmarkMultiGetSetBlock_256_Shard-12 21827272 554.8 ns/op 64 B/op 2 allocs/op
BenchmarkKeys-12 12793 913201 ns/op 329755 B/op 70 allocs/op
PASS
ok github.com/orcaman/concurrent-map/v2 532.335s
Test result
> go clean -testcache
> go test ./...
ok github.com/orcaman/concurrent-map/v2 0.261s
Agreed. This is a good feature to have and I'm currently waiting for this to be released too. Hope maintainers have no issues with this.
