mirror of
https://github.com/golang/go.git
synced 2025-05-06 08:03:03 +00:00
6f9b94ab66
This CL implements the math/bits.OnesCount{8,16,32,64} functions
as intrinsics on s390x using the 'population count' (popcnt)
instruction. This instruction was released as the 'population-count'
facility which uses the same facility bit (45) as the
'distinct-operands' facility which is a pre-requisite for Go on
s390x. We can therefore use it without a feature check.
The s390x popcnt instruction treats a 64 bit register as a vector
of 8 bytes, summing the number of ones in each byte individually.
It then writes the results to the corresponding bytes in the
output register. Therefore to implement OnesCount{16,32,64} we
need to sum the individual byte counts using some extra
instructions. To do this efficiently I've added some additional
pseudo operations to the s390x SSA backend.
Unlike other architectures the new instruction sequence is faster
for OnesCount8, so that is implemented using the intrinsic.
name old time/op new time/op delta
OnesCount 3.21ns ± 1% 1.35ns ± 0% -58.00% (p=0.000 n=20+20)
OnesCount8 0.91ns ± 1% 0.81ns ± 0% -11.43% (p=0.000 n=20+20)
OnesCount16 1.51ns ± 3% 1.21ns ± 0% -19.71% (p=0.000 n=20+17)
OnesCount32 1.91ns ± 0% 1.12ns ± 1% -41.60% (p=0.000 n=19+20)
OnesCount64 3.18ns ± 4% 1.35ns ± 0% -57.52% (p=0.000 n=20+20)
Change-Id: Id54f0bd28b6db9a887ad12c0d72fcc168ef9c4e0
Reviewed-on: https://go-review.googlesource.com/114675
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
279 lines
5.1 KiB
Go
279 lines
5.1 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"
|
|
// s390x:"POPCNT"
|
|
return bits.OnesCount(n)
|
|
}
|
|
|
|
func OnesCount64(n uint64) int {
|
|
// amd64:"POPCNTQ",".*support_popcnt"
|
|
// arm64:"VCNT","VUADDLV"
|
|
// s390x:"POPCNT"
|
|
return bits.OnesCount64(n)
|
|
}
|
|
|
|
func OnesCount32(n uint32) int {
|
|
// amd64:"POPCNTL",".*support_popcnt"
|
|
// arm64:"VCNT","VUADDLV"
|
|
// s390x:"POPCNT"
|
|
return bits.OnesCount32(n)
|
|
}
|
|
|
|
func OnesCount16(n uint16) int {
|
|
// amd64:"POPCNTL",".*support_popcnt"
|
|
// arm64:"VCNT","VUADDLV"
|
|
// s390x:"POPCNT"
|
|
return bits.OnesCount16(n)
|
|
}
|
|
|
|
func OnesCount8(n uint8) int {
|
|
// s390x:"POPCNT"
|
|
return bits.OnesCount8(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
|
|
}
|