encoding/{base32,base64}: speed up Encode

This CL clarifies (*Encoding).Encode and speeds it up by reducing the
number of bounds checks in its loop.

Here are some benchmark results (no change to allocations):

goos: darwin
goarch: arm64
pkg: encoding/base32
cpu: Apple M4
                  │     old     │             new              │
                  │   sec/op    │   sec/op     vs base         │
EncodeToString-10   7.310µ ± 0%   5.308µ ± 0%  -27.39% (n=180)
Encode-10           5.651µ ± 0%   3.603µ ± 0%  -36.25% (n=180)
geomean             6.427µ        4.373µ       -31.96%

                  │     old      │                  new                  │
                  │     B/s      │     B/s       vs base                 │
EncodeToString-10   1.044Gi ± 0%   1.437Gi ± 0%  +37.71% (p=0.000 n=180)
Encode-10           1.350Gi ± 0%   2.118Gi ± 0%  +56.88% (p=0.000 n=180)
geomean             1.187Gi        1.745Gi       +46.98%

pkg: encoding/base64
                  │     old     │             new              │
                  │   sec/op    │   sec/op     vs base         │
EncodeToString-10   7.058µ ± 0%   6.034µ ± 0%  -14.51% (n=180)

                  │     old      │                  new                  │
                  │     B/s      │     B/s       vs base                 │
EncodeToString-10   1.081Gi ± 0%   1.264Gi ± 0%  +16.97% (p=0.000 n=180)

Updates #20206

Change-Id: I7d46891ddb4371df004bfd612a8efc6638715b94
GitHub-Last-Rev: 1caac3d65532fefacbbed57f11a4a49273f173e2
GitHub-Pull-Request: golang/go#78344
Reviewed-on: https://go-review.googlesource.com/c/go/+/759100
Reviewed-by: Keith Randall <khr@golang.org>
Auto-Submit: Keith Randall <khr@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Keith Randall <khr@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

1 files changed, 24 insertions, 29 deletions

diff --git a/src/encoding/base64/base64.go b/src/encoding/base64/base64.go
index 57aa1a697f..32014f45bb 100644
--- a/src/encoding/base64/base64.go
+++ b/src/encoding/base64/base64.go
@@ -151,44 +151,39 @@ func (enc *Encoding) Encode(dst, src []byte) {
 	// outside of the loop to speed up the encoder.
 	_ = enc.encode
 
-	di, si := 0, 0
-	n := (len(src) / 3) * 3
-	for si < n {
+	for len(src) >= 3 {
 		// Convert 3x 8bit source bytes into 4 bytes
-		val := uint(src[si+0])<<16 | uint(src[si+1])<<8 | uint(src[si+2])
+		val := uint(src[0])<<16 | uint(src[1])<<8 | uint(src[2])
 
-		dst[di+0] = enc.encode[val>>18&0x3F]
-		dst[di+1] = enc.encode[val>>12&0x3F]
-		dst[di+2] = enc.encode[val>>6&0x3F]
-		dst[di+3] = enc.encode[val&0x3F]
+		_ = dst[3] // Eliminate bounds checks below.
+		dst[0] = enc.encode[val>>18&0x3F]
+		dst[1] = enc.encode[val>>12&0x3F]
+		dst[2] = enc.encode[val>>6&0x3F]
+		dst[3] = enc.encode[val&0x3F]
 
-		si += 3
-		di += 4
+		src = src[3:]
+		dst = dst[4:]
 	}
 
-	remain := len(src) - si
-	if remain == 0 {
+	// Add the remaining small block (if any).
+	switch len(src) {
+	case 0:
 		return
-	}
-	// Add the remaining small block
-	val := uint(src[si+0]) << 16
-	if remain == 2 {
-		val |= uint(src[si+1]) << 8
-	}
-
-	dst[di+0] = enc.encode[val>>18&0x3F]
-	dst[di+1] = enc.encode[val>>12&0x3F]
-
-	switch remain {
-	case 2:
-		dst[di+2] = enc.encode[val>>6&0x3F]
+	case 1:
+		val := uint(src[0]) << 16
+		dst[0] = enc.encode[val>>18&0x3F]
+		dst[1] = enc.encode[val>>12&0x3F]
 		if enc.padChar != NoPadding {
-			dst[di+3] = byte(enc.padChar)
+			dst[2] = byte(enc.padChar)
+			dst[3] = byte(enc.padChar)
 		}
-	case 1:
+	case 2:
+		val := uint(src[0])<<16 | uint(src[1])<<8
+		dst[0] = enc.encode[val>>18&0x3F]
+		dst[1] = enc.encode[val>>12&0x3F]
+		dst[2] = enc.encode[val>>6&0x3F]
 		if enc.padChar != NoPadding {
-			dst[di+2] = byte(enc.padChar)
-			dst[di+3] = byte(enc.padChar)
+			dst[3] = byte(enc.padChar)
 		}
 	}
 }