cmd/compile: optimize 386's math.bits.TrailingZeros16

This CL optimizes math.bits.TrailingZeros16 on 386 with a pair of BSFL and ORL instrcutions. The case TrailingZeros16-4 of the benchmark test in math/bits shows big improvement. name old time/op new time/op delta TrailingZeros16-4 1.55ns ± 1% 0.87ns ± 1% -43.87% (p=0.000 n=50+49) Change-Id: Ia899975b0e46f45dcd20223b713ed632bc32740b Reviewed-on: https://go-review.googlesource.com/c/go/+/189277 Run-TryBot: Ben Shi <[email protected]> TryBot-Result: Gobot Gobot <[email protected]> Reviewed-by: Keith Randall <[email protected]>
cloudflare · Aug 28, 2019 · 22355d6 · 22355d6
1 parent e1e4c49
commit 22355d6
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 1 deletion.
diff --git a/src/cmd/compile/internal/gc/ssa.go b/src/cmd/compile/internal/gc/ssa.go
@@ -3370,7 +3370,7 @@ func init() {
  func(s *state, n *Node, args []*ssa.Value) *ssa.Value {
  return s.newValue1(ssa.OpCtz16, types.Types[TINT], args[0])
  },
- sys.AMD64, sys.ARM, sys.ARM64, sys.Wasm)
+ sys.AMD64, sys.ARM, sys.I386, sys.ARM64, sys.Wasm)
  addF("math/bits", "TrailingZeros16",
  func(s *state, n *Node, args []*ssa.Value) *ssa.Value {
  x := s.newValue1(ssa.OpZeroExt16to64, types.Types[TUINT64], args[0])

diff --git a/src/cmd/compile/internal/ssa/gen/386.rules b/src/cmd/compile/internal/ssa/gen/386.rules
@@ -66,6 +66,8 @@
 
 (Sqrt x) -> (SQRTSD x)
 
+(Ctz16 x) -> (BSFL (ORLconst <typ.UInt32> [0x10000] x))
+
 // Lowering extension
 (SignExt8to16 x) -> (MOVBLSX x)
 (SignExt8to32 x) -> (MOVBLSX x)

diff --git a/src/cmd/compile/internal/ssa/rewrite386.go b/src/cmd/compile/internal/ssa/rewrite386.go
diff --git a/test/codegen/mathbits.go b/test/codegen/mathbits.go
@@ -296,6 +296,7 @@ func TrailingZeros32(n uint32) int {
 
 func TrailingZeros16(n uint16) int {
  // amd64:"BSFL","BTSL\\t\\$16"
+ // 386:"BSFL\t"
  // arm:"ORR\t\\$65536","CLZ",-"MOVHU\tR"
  // arm64:"ORR\t\\$65536","RBITW","CLZW",-"MOVHU\tR",-"RBIT\t",-"CLZ\t"
  // s390x:"FLOGR","OR\t\\$65536"