improving math/bits speed

Author: grantnelson-wf

compiler/expressions.go

@@ -667,6 +667,12 @@ func (fc *funcContext) translateExpr(expr ast.Expr) *expression {
 						return fc.formatExpr("debugger")
 					case "InternalObject":
 						return fc.translateExpr(e.Args[0])
+					case "MakeUint64":
+						return fc.formatExpr("new $Uint64(%e, %e)", e.Args[0], e.Args[1])
+					case "Uint64High":
+						return fc.formatExpr("%e.$high", e.Args[0])
+					case "Uint64Low":
+						return fc.formatExpr("%e.$low", e.Args[0])
 					}
 				}
 				return fc.translateCall(e, sig, fc.translateExpr(f))

compiler/natives/src/crypto/internal/bigmod/nat.go

@@ -0,0 +1,46 @@
+//go:build js
+
+package bigmod
+
+import "math/bits"
+
+// GOPHERJS: The original has specialized cases for 1024, 1536, and 2048 bit
+// sizes that use addMulVVW1024/1536/2048 which take *uint pointers.
+// Those functions then use unsafe.Slice to convert back to slices.
+// In GopherJS, creating pointers via &slice[i] generates an $indexPtr.
+// We avoid this by always using the no-asm slice-based implementation
+// which calls addMulVVW with slices directly.
+//
+//gopherjs:replace
+func (x *Nat) montgomeryMul(a *Nat, b *Nat, m *Modulus) *Nat {
+	n := len(m.nat.limbs)
+	mLimbs := m.nat.limbs[:n]
+	aLimbs := a.limbs[:n]
+	bLimbs := b.limbs[:n]
+
+	T := make([]uint, n*2)
+
+	var c uint
+	for i := 0; i < n; i++ {
+		_ = T[n+i] // bounds check elimination hint
+		d := bLimbs[i]
+		c1 := addMulVVW(T[i:n+i], aLimbs, d)
+		Y := T[i] * m.m0inv
+		c2 := addMulVVW(T[i:n+i], mLimbs, Y)
+		T[n+i], c = bits.Add(c1, c2, c)
+	}
+
+	copy(x.reset(n).limbs, T[n:])
+	x.maybeSubtractModulus(choice(c), m)
+
+	return x
+}
+
+//gopherjs:purge
+func addMulVVW1024(z, x *uint, y uint) (c uint)
+
+//gopherjs:purge
+func addMulVVW1536(z, x *uint, y uint) (c uint)
+
+//gopherjs:purge
+func addMulVVW2048(z, x *uint, y uint) (c uint)