"Drain Random from Chaos. Breeze - a new family of fast CB-PRNG"
Actual version of the breeze PRNG: (That was the old one with many blabla: https://github.com/AndreasBriese/breeze)
import with go get github.com/AndreasBriese/breeze32
2018 Easier structure and even faster and more reliable.
As long as breeze32 is running on one cpu, it appeared to be threadsafe even with intensive use.
If you need a multiprocessor safe prng use the ByteMP function (and/or adapt it for the other variants). And consider applying the mutex in roundTrip().
See the test file for usage.
Benchmarks (go test -bench=.)
goos: darwin
goarch: amd64
pkg: github.com/AndreasBriese/breeze32
Benchmark_Init_and_Seed 10000000 208 ns/op
Benchmark_breeze_Uint64 200000000 8.92 ns/op
Benchmark_breeze_Uint32 200000000 7.32 ns/op
Benchmark_breeze_Uint8 200000000 6.42 ns/op
Benchmark_breeze_ByteMP 20000000 64.5 ns/op
Benchmark_breeze_RandIntN_8byte 100000000 18.8 ns/op
Benchmark_breeze_RandIntN_16byte 100000000 18.4 ns/op
Benchmark_breeze_RandIntN_32byte 100000000 21.0 ns/op
Benchmark_breeze_RandIntN_64byte 50000000 22.9 ns/op
Benchmark_breeze_Float64 100000000 12.5 ns/op
Benchmark_breeze_Float32 100000000 11.2 ns/op
Benchmark_mathRand_Intn_8byte 100000000 17.1 ns/op
Benchmark_mathRand_Intn_16byte 100000000 17.1 ns/op
Benchmark_mathRand_Intn_32byte 100000000 14.8 ns/op
Benchmark_mathRand_Intn_64byte 30000000 36.6 ns/op
Benchmark_mathRand_Float_64 100000000 13.0 ns/op
Benchmark_mathRand_Float_32 100000000 16.1 ns/op
PASS
ok github.com/AndreasBriese/breeze32 30.911s
go version go1.10.2 darwin/amd64 on MBP2013 i7 8GB Ram
License MIT-Style for software use // see License
Dr. Andreas Briese
2018/09/24
eduToolbox@Bri-C GmbH, Sarstedt