golang/go
0
1
Fork 0
mirror of https://github.com/golang/go synced 2025-06-15 17:41:45 +00:00
Code Packages Activity

cmd/compile: combine multiply/add into maddld on ppc64le/power9

Browse Source 
Add a new lowering rule to match and replace such instances
with the MADDLD instruction available on power9 where
possible.

Likewise, this plumbs in a new ppc64 ssa opcode to house
the newly generated MADDLD instructions.

When testing ed25519, this reduced binary size by 936B.
Similarly, MADDLD combination occcurs in a few other less
obvious cases such as division by constant.

Testing of golang.org/x/crypto/ed25519 shows non-trivial
speedup during keygeneration:

name           old time/op  new time/op  delta
KeyGeneration  65.2µs ± 0%  63.1µs ± 0%  -3.19%
Signing        64.3µs ± 0%  64.4µs ± 0%  +0.16%
Verification    147µs ± 0%   147µs ± 0%  +0.11%

Similarly, this test binary has shrunk by 66488B.

Change-Id: I077aeda7943119b41f07e4e62e44a648f16e4ad0
Reviewed-on: https://go-review.googlesource.com/c/go/+/248723
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
This commit is contained in:
Paul E. Murphy
2020-08-17 16:14:48 -05:00
committed by Lynn Boger
parent 216714e44f
commit e7c7ce646f
6 changed files with 64 additions and 4 deletions
src/cmd/compile/internal
ppc64
ssa.go
ssa
gen
PPC64.rulesPPC64Ops.go
opGen.gorewritePPC64.go
test/codegen
arithmetic.go

12
test/codegen/arithmetic.go

@ -253,16 +253,20 @@ func Divisible(n1 uint, n2 int) (bool, bool, bool, bool) {
// 386:"IMUL3L\t[$]-1431655765","ADDL\t[$]715827882","ROLL\t[$]31",-"DIVQ"
// arm64:"MUL","ADD\t[$]3074457345618258602","ROR",-"DIV"
// arm:"MUL","ADD\t[$]715827882",-".*udiv"
// ppc64:"MULLD","ADD","ROTL\t[$]63"
// ppc64le:"MULLD","ADD","ROTL\t[$]63"
// ppc64/power8:"MULLD","ADD","ROTL\t[$]63"
// ppc64le/power8:"MULLD","ADD","ROTL\t[$]63"
// ppc64/power9:"MADDLD","ROTL\t[$]63"
// ppc64le/power9:"MADDLD","ROTL\t[$]63"
evenS := n2%6 == 0
// amd64:"IMULQ","ADD",-"ROLQ",-"DIVQ"
// 386:"IMUL3L\t[$]678152731","ADDL\t[$]113025455",-"ROLL",-"DIVQ"
// arm64:"MUL","ADD\t[$]485440633518672410",-"ROR",-"DIV"
// arm:"MUL","ADD\t[$]113025455",-".*udiv"
// ppc64:"MULLD","ADD",-"ROTL"
// ppc64le:"MULLD","ADD",-"ROTL"
// ppc64/power8:"MULLD","ADD",-"ROTL"
// ppc64/power9:"MADDLD",-"ROTL"
// ppc64le/power8:"MULLD","ADD",-"ROTL"
// ppc64le/power9:"MADDLD",-"ROTL"
oddS := n2%19 == 0
return evenU, oddU, evenS, oddS