mirror of
https://github.com/golang/go.git
synced 2025-10-25 05:53:20 +00:00
cmd/compile: add math/bits.{Add,Sub}64 intrinsics on s390x
This CL adds intrinsics for the 64-bit addition and subtraction functions in math/bits. These intrinsics use the condition code to propagate the carry or borrow bit. To make the carry chains more efficient I've removed the 'clobberFlags' property from most of the load and store operations. Originally these ops did clobber flags when using offsets that didn't fit in a signed 20-bit integer, however that is no longer true. As with other platforms the intrinsics are faster when executed in a chain rather than a loop because currently we need to spill and restore the carry bit between each loop iteration. We may be able to reduce the need to do this on s390x (e.g. by using compare-and-branch instructions that do not clobber flags) in the future. name old time/op new time/op delta Add64 1.21ns ± 2% 2.03ns ± 2% +67.18% (p=0.000 n=7+10) Add64multiple 2.98ns ± 3% 1.03ns ± 0% -65.39% (p=0.000 n=10+9) Sub64 1.23ns ± 4% 2.03ns ± 1% +64.85% (p=0.000 n=10+10) Sub64multiple 3.73ns ± 4% 1.04ns ± 1% -72.28% (p=0.000 n=10+8) Change-Id: I913bbd5e19e6b95bef52f5bc4f14d6fe40119083 Reviewed-on: https://go-review.googlesource.com/c/go/+/174303 Run-TryBot: Michael Munday <mike.munday@ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
This commit is contained in:
Michael Munday
parent
004fb5cb8d
commit
2c1b5130aa
9 changed files with 880 additions and 195 deletions
|
|
@ -377,32 +377,38 @@ func IterateBits8(n uint8) int {
|
|||
func Add(x, y, ci uint) (r, co uint) {
|
||||
// arm64:"ADDS","ADCS","ADC",-"ADD\t",-"CMP"
|
||||
// amd64:"NEGL","ADCQ","SBBQ","NEGQ"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
return bits.Add(x, y, ci)
|
||||
}
|
||||
|
||||
func AddC(x, ci uint) (r, co uint) {
|
||||
// arm64:"ADDS","ADCS","ADC",-"ADD\t",-"CMP"
|
||||
// amd64:"NEGL","ADCQ","SBBQ","NEGQ"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
return bits.Add(x, 7, ci)
|
||||
}
|
||||
|
||||
func AddZ(x, y uint) (r, co uint) {
|
||||
// arm64:"ADDS","ADC",-"ADCS",-"ADD\t",-"CMP"
|
||||
// amd64:"ADDQ","SBBQ","NEGQ",-"NEGL",-"ADCQ"
|
||||
// s390x:"ADDC",-"ADDC\t[$]-1,"
|
||||
return bits.Add(x, y, 0)
|
||||
}
|
||||
|
||||
func AddR(x, y, ci uint) uint {
|
||||
// arm64:"ADDS","ADCS",-"ADD\t",-"CMP"
|
||||
// amd64:"NEGL","ADCQ",-"SBBQ",-"NEGQ"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
r, _ := bits.Add(x, y, ci)
|
||||
return r
|
||||
}
|
||||
|
||||
func AddM(p, q, r *[3]uint) {
|
||||
var c uint
|
||||
r[0], c = bits.Add(p[0], q[0], c)
|
||||
// arm64:"ADCS",-"ADD\t",-"CMP"
|
||||
// amd64:"ADCQ",-"NEGL",-"SBBQ",-"NEGQ"
|
||||
// s390x:"ADDE",-"ADDC\t[$]-1,"
|
||||
r[1], c = bits.Add(p[1], q[1], c)
|
||||
r[2], c = bits.Add(p[2], q[2], c)
|
||||
}
|
||||
|
|
@ -412,6 +418,7 @@ func Add64(x, y, ci uint64) (r, co uint64) {
|
|||
// amd64:"NEGL","ADCQ","SBBQ","NEGQ"
|
||||
// ppc64: "ADDC", "ADDE", "ADDZE"
|
||||
// ppc64le: "ADDC", "ADDE", "ADDZE"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
return bits.Add64(x, y, ci)
|
||||
}
|
||||
|
||||
|
|
@ -420,6 +427,7 @@ func Add64C(x, ci uint64) (r, co uint64) {
|
|||
// amd64:"NEGL","ADCQ","SBBQ","NEGQ"
|
||||
// ppc64: "ADDC", "ADDE", "ADDZE"
|
||||
// ppc64le: "ADDC", "ADDE", "ADDZE"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
return bits.Add64(x, 7, ci)
|
||||
}
|
||||
|
||||
|
|
@ -428,6 +436,7 @@ func Add64Z(x, y uint64) (r, co uint64) {
|
|||
// amd64:"ADDQ","SBBQ","NEGQ",-"NEGL",-"ADCQ"
|
||||
// ppc64: "ADDC", "ADDE", "ADDZE"
|
||||
// ppc64le: "ADDC", "ADDE", "ADDZE"
|
||||
// s390x:"ADDC",-"ADDC\t[$]-1,"
|
||||
return bits.Add64(x, y, 0)
|
||||
}
|
||||
|
||||
|
|
@ -436,6 +445,7 @@ func Add64R(x, y, ci uint64) uint64 {
|
|||
// amd64:"NEGL","ADCQ",-"SBBQ",-"NEGQ"
|
||||
// ppc64: "ADDC", "ADDE", "ADDZE"
|
||||
// ppc64le: "ADDC", "ADDE", "ADDZE"
|
||||
// s390x:"ADDE","ADDC\t[$]-1,"
|
||||
r, _ := bits.Add64(x, y, ci)
|
||||
return r
|
||||
}
|
||||
|
|
@ -446,6 +456,7 @@ func Add64M(p, q, r *[3]uint64) {
|
|||
// amd64:"ADCQ",-"NEGL",-"SBBQ",-"NEGQ"
|
||||
// ppc64: "ADDC", "ADDE", "ADDZE"
|
||||
// ppc64le: "ADDC", "ADDE", "ADDZE"
|
||||
// s390x:"ADDE",-"ADDC\t[$]-1,"
|
||||
r[1], c = bits.Add64(p[1], q[1], c)
|
||||
r[2], c = bits.Add64(p[2], q[2], c)
|
||||
}
|
||||
|
|
@ -457,24 +468,28 @@ func Add64M(p, q, r *[3]uint64) {
|
|||
func Sub(x, y, ci uint) (r, co uint) {
|
||||
// amd64:"NEGL","SBBQ","NEGQ"
|
||||
// arm64:"NEGS","SBCS","NGC","NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
return bits.Sub(x, y, ci)
|
||||
}
|
||||
|
||||
func SubC(x, ci uint) (r, co uint) {
|
||||
// amd64:"NEGL","SBBQ","NEGQ"
|
||||
// arm64:"NEGS","SBCS","NGC","NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
return bits.Sub(x, 7, ci)
|
||||
}
|
||||
|
||||
func SubZ(x, y uint) (r, co uint) {
|
||||
// amd64:"SUBQ","SBBQ","NEGQ",-"NEGL"
|
||||
// arm64:"SUBS","NGC","NEG",-"SBCS",-"ADD",-"SUB\t",-"CMP"
|
||||
// s390x:"SUBC"
|
||||
return bits.Sub(x, y, 0)
|
||||
}
|
||||
|
||||
func SubR(x, y, ci uint) uint {
|
||||
// amd64:"NEGL","SBBQ",-"NEGQ"
|
||||
// arm64:"NEGS","SBCS",-"NGC",-"NEG\t",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
r, _ := bits.Sub(x, y, ci)
|
||||
return r
|
||||
}
|
||||
|
|
@ -483,6 +498,7 @@ func SubM(p, q, r *[3]uint) {
|
|||
r[0], c = bits.Sub(p[0], q[0], c)
|
||||
// amd64:"SBBQ",-"NEGL",-"NEGQ"
|
||||
// arm64:"SBCS",-"NEGS",-"NGC",-"NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
r[1], c = bits.Sub(p[1], q[1], c)
|
||||
r[2], c = bits.Sub(p[2], q[2], c)
|
||||
}
|
||||
|
|
@ -490,24 +506,28 @@ func SubM(p, q, r *[3]uint) {
|
|||
func Sub64(x, y, ci uint64) (r, co uint64) {
|
||||
// amd64:"NEGL","SBBQ","NEGQ"
|
||||
// arm64:"NEGS","SBCS","NGC","NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
return bits.Sub64(x, y, ci)
|
||||
}
|
||||
|
||||
func Sub64C(x, ci uint64) (r, co uint64) {
|
||||
// amd64:"NEGL","SBBQ","NEGQ"
|
||||
// arm64:"NEGS","SBCS","NGC","NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
return bits.Sub64(x, 7, ci)
|
||||
}
|
||||
|
||||
func Sub64Z(x, y uint64) (r, co uint64) {
|
||||
// amd64:"SUBQ","SBBQ","NEGQ",-"NEGL"
|
||||
// arm64:"SUBS","NGC","NEG",-"SBCS",-"ADD",-"SUB\t",-"CMP"
|
||||
// s390x:"SUBC"
|
||||
return bits.Sub64(x, y, 0)
|
||||
}
|
||||
|
||||
func Sub64R(x, y, ci uint64) uint64 {
|
||||
// amd64:"NEGL","SBBQ",-"NEGQ"
|
||||
// arm64:"NEGS","SBCS",-"NGC",-"NEG\t",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
r, _ := bits.Sub64(x, y, ci)
|
||||
return r
|
||||
}
|
||||
|
|
@ -516,6 +536,7 @@ func Sub64M(p, q, r *[3]uint64) {
|
|||
r[0], c = bits.Sub64(p[0], q[0], c)
|
||||
// amd64:"SBBQ",-"NEGL",-"NEGQ"
|
||||
// arm64:"SBCS",-"NEGS",-"NGC",-"NEG",-"ADD",-"SUB",-"CMP"
|
||||
// s390x:"SUBE"
|
||||
r[1], c = bits.Sub64(p[1], q[1], c)
|
||||
r[2], c = bits.Sub64(p[2], q[2], c)
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue