core/crypto/chacha20: Misc Simd128 improvements

- Detect the RISC-V `v` profile
- Don't bother trying to process 4 blocks at a time if emulated
This commit is contained in:
Yawning Angel
2024-08-28 04:11:55 +09:00
parent e4e76f27f6
commit 6e8710fce4

View File

@@ -29,11 +29,24 @@ when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 {
// explicitly using simd.u8x16 shuffles.
@(private = "file")
TARGET_SIMD_FEATURES :: "sse2,ssse3"
} else when ODIN_ARCH == .riscv64 {
@(private = "file")
TARGET_SIMD_FEATURES :: "v"
} else {
@(private = "file")
TARGET_SIMD_FEATURES :: ""
}
// Some targets lack runtime feature detection, and will flat out refuse
// to load binaries that have unknown instructions. This is distinct from
// `simd.IS_EMULATED` as actually good designs support runtime feature
// detection and that constant establishes a baseline.
//
// See:
// - https://github.com/WebAssembly/design/issues/1161
@(private = "file")
TARGET_IS_DESIGNED_BY_IDIOTS :: (ODIN_ARCH == .wasm64p32 || ODIN_ARCH == .wasm32) && !intrinsics.has_target_feature("simd128")
@(private = "file")
_ROT_7L: simd.u32x4 : {7, 7, 7, 7}
@(private = "file")
@@ -205,11 +218,13 @@ _store_simd128 :: #force_inline proc "contextless" (
// is_performant returns true iff the target and current host both support
// "enough" 128-bit SIMD to make this implementation performant.
is_performant :: proc "contextless" () -> bool {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 || ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 || ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 || ODIN_ARCH == .riscv64 {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 {
req_features :: info.CPU_Features{.asimd}
} else when ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 {
req_features :: info.CPU_Features{.sse2, .ssse3}
} else when ODIN_ARCH == .riscv64 {
req_features :: info.CPU_Features{.V}
}
features, ok := info.cpu_features.?
@@ -245,8 +260,17 @@ stream_blocks :: proc(ctx: ^_chacha20.Context, dst, src: []byte, nr_blocks: int)
// 8 blocks at a time.
//
// Note: This is only worth it on Aarch64.
when ODIN_ARCH == .arm64 {
// Note:
// This uses a ton of registers so it is only worth it on targets
// that have something like 32 128-bit registers. This is currently
// all ARMv8 targets, and RISC-V Zvl128b (`V` application profile)
// targets.
//
// While our current definition of `.arm32` is 32-bit ARMv8, this
// may change in the future (ARMv7 is still relevant), and things
// like Cortex-A8/A9 does "pretend" 128-bit SIMD 64-bits at a time
// thus needs bemchmarking.
when ODIN_ARCH == .arm64 || ODIN_ARCH == .riscv64 {
for ; n >= 8; n = n - 8 {
v0, v1, v2, v3 := s0, s1, s2, s3
@@ -354,9 +378,11 @@ stream_blocks :: proc(ctx: ^_chacha20.Context, dst, src: []byte, nr_blocks: int)
// 4 blocks at a time.
//
// Note: The i386 target lacks the required number of registers
// for this to be performant, so it is skipped.
when ODIN_ARCH != .i386 {
// Note: This is skipped on several targets for various reasons.
// - i386 lacks the required number of registers
// - Generating code when runtime "hardware" SIMD support is impossible
// to detect is pointless, since this will be emulated using GP regs.
when ODIN_ARCH != .i386 && !TARGET_IS_DESIGNED_BY_IDIOTS {
for ; n >= 4; n = n - 4 {
v0, v1, v2, v3 := s0, s1, s2, s3