1// Copyright 2023 The Go Authors. All rights reserved.
2// Use of this source code is governed by a BSD-style
3// license that can be found in the LICENSE file.
4
5//go:build !purego && (386 || amd64 || arm || arm64 || ppc64 || ppc64le || riscv64 || s390x)
6
7package bigmod
8
9import "internal/cpu"
10
11// amd64 assembly uses ADCX/ADOX/MULX if ADX is available to run two carry
12// chains in the flags in parallel across the whole operation, and aggressively
13// unrolls loops. arm64 processes four words at a time.
14//
15// It's unclear why the assembly for all other architectures, as well as for
16// amd64 without ADX, perform better than the compiler output.
17// TODO(filippo): file cmd/compile performance issue.
18
19var supportADX = cpu.X86.HasADX && cpu.X86.HasBMI2
20
21//go:noescape
22func addMulVVW1024(z, x *uint, y uint) (c uint)
23
24//go:noescape
25func addMulVVW1536(z, x *uint, y uint) (c uint)
26
27//go:noescape
28func addMulVVW2048(z, x *uint, y uint) (c uint)
29