go/test/codegen/mathbits.go
Josh Bleecher Snyder d9a50a6531 cmd/compile: use prove pass to detect Ctz of non-zero values
On amd64, Ctz must include special handling of zeros.
But the prove pass has enough information to detect whether the input
is non-zero, allowing a more efficient lowering.

Introduce new CtzNonZero ops to capture and use this information.

Benchmark code:

func BenchmarkVisitBits(b *testing.B) {
	b.Run("8", func(b *testing.B) {
		for i := 0; i < b.N; i++ {
			x := uint8(0xff)
			for x != 0 {
				sink = bits.TrailingZeros8(x)
				x &= x - 1
			}
		}
	})

    // and similarly so for 16, 32, 64
}

name            old time/op  new time/op  delta
VisitBits/8-8   7.27ns ± 4%  5.58ns ± 4%  -23.35%  (p=0.000 n=28+26)
VisitBits/16-8  14.7ns ± 7%  10.5ns ± 4%  -28.43%  (p=0.000 n=30+28)
VisitBits/32-8  27.6ns ± 8%  19.3ns ± 3%  -30.14%  (p=0.000 n=30+26)
VisitBits/64-8  44.0ns ±11%  38.0ns ± 5%  -13.48%  (p=0.000 n=30+30)

Fixes #25077

Change-Id: Ie6e5bd86baf39ee8a4ca7cadcf56d934e047f957
Reviewed-on: https://go-review.googlesource.com/109358
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
2018-04-26 18:22:28 +00:00

270 lines
5 KiB
Go

// asmcheck
// Copyright 2018 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package codegen
import "math/bits"
// ----------------------- //
// bits.LeadingZeros //
// ----------------------- //
func LeadingZeros(n uint) int {
// amd64:"BSRQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.LeadingZeros(n)
}
func LeadingZeros64(n uint64) int {
// amd64:"BSRQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.LeadingZeros64(n)
}
func LeadingZeros32(n uint32) int {
// amd64:"BSRQ","LEAQ",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.LeadingZeros32(n)
}
func LeadingZeros16(n uint16) int {
// amd64:"BSRL","LEAL",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.LeadingZeros16(n)
}
func LeadingZeros8(n uint8) int {
// amd64:"BSRL","LEAL",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.LeadingZeros8(n)
}
// --------------- //
// bits.Len* //
// --------------- //
func Len(n uint) int {
// amd64:"BSRQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.Len(n)
}
func Len64(n uint64) int {
// amd64:"BSRQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.Len64(n)
}
func Len32(n uint32) int {
// amd64:"BSRQ","LEAQ",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.Len32(n)
}
func Len16(n uint16) int {
// amd64:"BSRL","LEAL",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.Len16(n)
}
func Len8(n uint8) int {
// amd64:"BSRL","LEAL",-"CMOVQEQ"
// s390x:"FLOGR"
// arm:"CLZ" arm64:"CLZ"
// mips:"CLZ"
return bits.Len8(n)
}
// -------------------- //
// bits.OnesCount //
// -------------------- //
func OnesCount(n uint) int {
// amd64:"POPCNTQ",".*support_popcnt"
// arm64:"VCNT","VUADDLV"
return bits.OnesCount(n)
}
func OnesCount64(n uint64) int {
// amd64:"POPCNTQ",".*support_popcnt"
// arm64:"VCNT","VUADDLV"
return bits.OnesCount64(n)
}
func OnesCount32(n uint32) int {
// amd64:"POPCNTL",".*support_popcnt"
// arm64:"VCNT","VUADDLV"
return bits.OnesCount32(n)
}
func OnesCount16(n uint16) int {
// amd64:"POPCNTL",".*support_popcnt"
// arm64:"VCNT","VUADDLV"
return bits.OnesCount16(n)
}
// ----------------------- //
// bits.ReverseBytes //
// ----------------------- //
func ReverseBytes(n uint) uint {
// amd64:"BSWAPQ"
// s390x:"MOVDBR"
// arm64:"REV"
return bits.ReverseBytes(n)
}
func ReverseBytes64(n uint64) uint64 {
// amd64:"BSWAPQ"
// s390x:"MOVDBR"
// arm64:"REV"
return bits.ReverseBytes64(n)
}
func ReverseBytes32(n uint32) uint32 {
// amd64:"BSWAPL"
// s390x:"MOVWBR"
// arm64:"REVW"
return bits.ReverseBytes32(n)
}
func ReverseBytes16(n uint16) uint16 {
// amd64:"ROLW"
return bits.ReverseBytes16(n)
}
// --------------------- //
// bits.RotateLeft //
// --------------------- //
func RotateLeft64(n uint64) uint64 {
// amd64:"ROLQ"
// arm64:"ROR"
// ppc64:"ROTL"
return bits.RotateLeft64(n, 37)
}
func RotateLeft32(n uint32) uint32 {
// amd64:"ROLL" 386:"ROLL"
// arm64:"RORW"
// ppc64:"ROTLW"
return bits.RotateLeft32(n, 9)
}
func RotateLeft16(n uint16) uint16 {
// amd64:"ROLW" 386:"ROLW"
return bits.RotateLeft16(n, 5)
}
func RotateLeft8(n uint8) uint8 {
// amd64:"ROLB" 386:"ROLB"
return bits.RotateLeft8(n, 5)
}
// ------------------------ //
// bits.TrailingZeros //
// ------------------------ //
func TrailingZeros(n uint) int {
// amd64:"BSFQ","MOVL\t\\$64","CMOVQEQ"
// s390x:"FLOGR"
return bits.TrailingZeros(n)
}
func TrailingZeros64(n uint64) int {
// amd64:"BSFQ","MOVL\t\\$64","CMOVQEQ"
// s390x:"FLOGR"
return bits.TrailingZeros64(n)
}
func TrailingZeros32(n uint32) int {
// amd64:"BTSQ\\t\\$32","BSFQ"
// s390x:"FLOGR","MOVWZ"
return bits.TrailingZeros32(n)
}
func TrailingZeros16(n uint16) int {
// amd64:"BSFL","BTSL\\t\\$16"
// s390x:"FLOGR","OR\t\\$65536"
return bits.TrailingZeros16(n)
}
func TrailingZeros8(n uint8) int {
// amd64:"BSFL","BTSL\\t\\$8"
// s390x:"FLOGR","OR\t\\$256"
return bits.TrailingZeros8(n)
}
// IterateBitsNN checks special handling of TrailingZerosNN when the input is known to be non-zero.
func IterateBits(n uint) int {
i := 0
for n != 0 {
// amd64:"BSFQ",-"CMOVEQ"
i += bits.TrailingZeros(n)
n &= n - 1
}
return i
}
func IterateBits64(n uint64) int {
i := 0
for n != 0 {
// amd64:"BSFQ",-"CMOVEQ"
i += bits.TrailingZeros64(n)
n &= n - 1
}
return i
}
func IterateBits32(n uint32) int {
i := 0
for n != 0 {
// amd64:"BSFL",-"BTSQ"
i += bits.TrailingZeros32(n)
n &= n - 1
}
return i
}
func IterateBits16(n uint16) int {
i := 0
for n != 0 {
// amd64:"BSFL",-"BTSL"
i += bits.TrailingZeros16(n)
n &= n - 1
}
return i
}
func IterateBits8(n uint8) int {
i := 0
for n != 0 {
// amd64:"BSFL",-"BTSL"
i += bits.TrailingZeros8(n)
n &= n - 1
}
return i
}