cmd/compile: optimize TrailingZeros(8|16) on amd64
Introduce Ctz8 and Ctz16 ops and provide optimized lowerings for them.
amd64 only for this CL, although it wouldn't surprise me
if other architectures also admit of optimized lowerings.
name old time/op new time/op delta
TrailingZeros8-8 1.33ns ± 6% 0.84ns ± 3% -36.90% (p=0.000 n=20+20)
TrailingZeros16-8 1.26ns ± 5% 0.84ns ± 5% -33.50% (p=0.000 n=20+18)
Code:
func f8(x uint8) { z = bits.TrailingZeros8(x) }
func f16(x uint16) { z = bits.TrailingZeros16(x) }
Before:
"".f8 STEXT nosplit size=34 args=0x8 locals=0x0
0x0000 00000 (x.go:7) TEXT "".f8(SB), NOSPLIT, $0-8
0x0000 00000 (x.go:7) FUNCDATA $0, gclocals·
2a5305abe05176240e61b8620e19a815(SB)
0x0000 00000 (x.go:7) FUNCDATA $1, gclocals·
33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0000 00000 (x.go:7) MOVBLZX "".x+8(SP), AX
0x0005 00005 (x.go:7) MOVBLZX AL, AX
0x0008 00008 (x.go:7) BTSQ $8, AX
0x000d 00013 (x.go:7) BSFQ AX, AX
0x0011 00017 (x.go:7) MOVL $64, CX
0x0016 00022 (x.go:7) CMOVQEQ CX, AX
0x001a 00026 (x.go:7) MOVQ AX, "".z(SB)
0x0021 00033 (x.go:7) RET
"".f16 STEXT nosplit size=34 args=0x8 locals=0x0
0x0000 00000 (x.go:8) TEXT "".f16(SB), NOSPLIT, $0-8
0x0000 00000 (x.go:8) FUNCDATA $0, gclocals·
2a5305abe05176240e61b8620e19a815(SB)
0x0000 00000 (x.go:8) FUNCDATA $1, gclocals·
33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0000 00000 (x.go:8) MOVWLZX "".x+8(SP), AX
0x0005 00005 (x.go:8) MOVWLZX AX, AX
0x0008 00008 (x.go:8) BTSQ $16, AX
0x000d 00013 (x.go:8) BSFQ AX, AX
0x0011 00017 (x.go:8) MOVL $64, CX
0x0016 00022 (x.go:8) CMOVQEQ CX, AX
0x001a 00026 (x.go:8) MOVQ AX, "".z(SB)
0x0021 00033 (x.go:8) RET
After:
"".f8 STEXT nosplit size=20 args=0x8 locals=0x0
0x0000 00000 (x.go:7) TEXT "".f8(SB), NOSPLIT, $0-8
0x0000 00000 (x.go:7) FUNCDATA $0, gclocals·
2a5305abe05176240e61b8620e19a815(SB)
0x0000 00000 (x.go:7) FUNCDATA $1, gclocals·
33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0000 00000 (x.go:7) MOVBLZX "".x+8(SP), AX
0x0005 00005 (x.go:7) BTSL $8, AX
0x0009 00009 (x.go:7) BSFL AX, AX
0x000c 00012 (x.go:7) MOVQ AX, "".z(SB)
0x0013 00019 (x.go:7) RET
"".f16 STEXT nosplit size=20 args=0x8 locals=0x0
0x0000 00000 (x.go:8) TEXT "".f16(SB), NOSPLIT, $0-8
0x0000 00000 (x.go:8) FUNCDATA $0, gclocals·
2a5305abe05176240e61b8620e19a815(SB)
0x0000 00000 (x.go:8) FUNCDATA $1, gclocals·
33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0000 00000 (x.go:8) MOVWLZX "".x+8(SP), AX
0x0005 00005 (x.go:8) BTSL $16, AX
0x0009 00009 (x.go:8) BSFL AX, AX
0x000c 00012 (x.go:8) MOVQ AX, "".z(SB)
0x0013 00019 (x.go:8) RET
Change-Id: I0551e357348de2b724737d569afd6ac9f5c3aa11
Reviewed-on: https://go-review.googlesource.com/108940
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Giovanni Bajo <rasky@develer.com>
Reviewed-by: Keith Randall <khr@golang.org>