bytes, strings: optimize Split*

The relevant benchmark results on linux/amd64:

bytes:

SplitSingleByteSeparator-4   25.7ms ± 5%   9.1ms ± 4%  -64.40%  (p=0.000 n=10+10)
SplitMultiByteSeparator-4    13.8ms ±20%   4.3ms ± 8%  -69.23%  (p=0.000 n=10+10)
SplitNSingleByteSeparator-4  1.88µs ± 9%  0.88µs ± 4%  -53.25%  (p=0.000 n=10+10)
SplitNMultiByteSeparator-4   4.83µs ±10%  1.32µs ± 9%  -72.65%  (p=0.000 n=10+10)

strings:

name                         old time/op  new time/op  delta
SplitSingleByteSeparator-4   21.4ms ± 8%   8.5ms ± 5%  -60.19%  (p=0.000 n=10+10)
SplitMultiByteSeparator-4    13.2ms ± 9%   3.9ms ± 4%  -70.29%  (p=0.000 n=10+10)
SplitNSingleByteSeparator-4  1.54µs ± 5%  0.75µs ± 7%  -51.21%  (p=0.000 n=10+10)
SplitNMultiByteSeparator-4   3.57µs ± 8%  1.01µs ±11%  -71.76%  (p=0.000 n=10+10)

Fixes #18973

Change-Id: Ie4bc010c6cc389001e72eab530497c81e5b26f34
Reviewed-on: https://go-review.googlesource.com/36510
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/bytes/bytes_test.go

@@ -1432,6 +1432,59 @@ func BenchmarkTrimSpace(b *testing.B) {
 	}
 }
 
+func makeBenchInputHard() []byte {
+	tokens := [...]string{
+		"<a>", "<p>", "<b>", "<strong>",
+		"</a>", "</p>", "</b>", "</strong>",
+		"hello", "world",
+	}
+	x := make([]byte, 0, 1<<20)
+	for {
+		i := rand.Intn(len(tokens))
+		if len(x)+len(tokens[i]) >= 1<<20 {
+			break
+		}
+		x = append(x, tokens[i]...)
+	}
+	return x
+}
+
+var benchInputHard = makeBenchInputHard()
+
+func BenchmarkSplitEmptySeparator(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		Split(benchInputHard, nil)
+	}
+}
+
+func BenchmarkSplitSingleByteSeparator(b *testing.B) {
+	sep := []byte("/")
+	for i := 0; i < b.N; i++ {
+		Split(benchInputHard, sep)
+	}
+}
+
+func BenchmarkSplitMultiByteSeparator(b *testing.B) {
+	sep := []byte("hello")
+	for i := 0; i < b.N; i++ {
+		Split(benchInputHard, sep)
+	}
+}
+
+func BenchmarkSplitNSingleByteSeparator(b *testing.B) {
+	sep := []byte("/")
+	for i := 0; i < b.N; i++ {
+		SplitN(benchInputHard, sep, 10)
+	}
+}
+
+func BenchmarkSplitNMultiByteSeparator(b *testing.B) {
+	sep := []byte("hello")
+	for i := 0; i < b.N; i++ {
+		SplitN(benchInputHard, sep, 10)
+	}
+}
+
 func BenchmarkRepeat(b *testing.B) {
 	for i := 0; i < b.N; i++ {
 		Repeat([]byte("-"), 80)