package nettrace
import (
"testing"
)
// Our benchmark loop is similar to the one from golang.org/x/net/trace, so we
// can compare results.
func runBench(b *testing.B, events int) {
nTraces := (b.N + events + 1) / events
for range nTraces {
tr := New("bench", "test")
for j := range events {
tr.Printf("%d", j)
}
tr.Finish()
}
}
func BenchmarkTrace_2(b *testing.B) {
runBench(b, 2)
}
func BenchmarkTrace_10(b *testing.B) {
runBench(b, 10)
}
func BenchmarkTrace_100(b *testing.B) {
runBench(b, 100)
}
func BenchmarkTrace_1000(b *testing.B) {
runBench(b, 1000)
}
func BenchmarkTrace_10000(b *testing.B) {
runBench(b, 10000)
}
func BenchmarkNewAndFinish(b *testing.B) {
for b.Loop() {
tr := New("bench", "test")
tr.Finish()
}
}
func BenchmarkPrintf(b *testing.B) {
tr := New("bench", "test")
defer tr.Finish()
for b.Loop() {
// Keep this without any formatting, so we measure our code instead of
// the performance of fmt.Sprintf.
tr.Printf("this is printf")
}
}