Add support for core.simd, intel-intrinsics, or inlined assembly

[dd86k]

Having some form of acceleration would benefit everyone, which this module currently lacks.

Options:

1. core.simd -- Supported everywhere, I think.
2. intel-intrinsics DUB package -- Somewhat supports all compilers.
3. Inlined assembly -- If all fails, at least x86 users would benefit. But limited to AVX/AV2 and not SSE* at best (because DMD).

The plan is to try options 1 and 2 and see which yields the best results (through benchmark and godbolt).

NOTE: The reason this wasn't implemented at first is because this module was once a contestant to get into Phobos.

[dd86k]

Or I could do like the std.digest.sha package does and at least support SSSE3 (version USE_SSSE3 is used from D_InlineAsm_X86 and D_InlineAsm_X86_64). At least I already have experience using the inline assembler under DMD, GDC, and LDC.

[dd86k]

I don't really think this needs SIMD this much because when compiled with LDC or GDC, I get similar performance results compared to OpenSSL.

Test env:

• sha3-d 1.2.1 with a buffer of 64 KiB (in ddh 1.3).
  • All compiled with release-nobounds.
• openssl 1.1.1f with default settings.
• Processor: Intel Core i7-3700
• Machine: VirtualBox 6.1.30 with KVM and guest additions
• OS: Ubuntu 20.04 LTS (5.11.0-43-generic)

Results (input: pv -r /dev/urandom piped):

• openssl dgst -sha3-256: 111-127 MiB/s
• ddh sha3-256 with dmd 2.098.1: 32-34 MiB/s (worse than 2.090.1 which is more around the 42 MiB/s mark)
• ddh sha3-256 with gdc 10.3: 90-91 MiB/s (worse since I upgraded dmd?!)
• ddh sha3-256 with ldc 1.20.1: 119-124 MiB/s

[dd86k]

New test under Windows (pv from Cygwin, supporting /dev/urandom) evaluates sha3-d at 140 MiB/s and OpenSSL-Win64 3.0.1 at 232 MiB/s so yeah I do see the difference now.

[dd86k]

In any case, a version Sha3dUseSIMD or Sha3dUseIntrinsics should be provided. Selecting it should be manual, feels better this way, to me at least.

[dd86k]

Plan:

• Use XCKP's SSE, AVX, and AVX2 implementation variants.
• version (D_AVX): Use AVX impl.
• version (D_AVX2): Use AVX2 impl.

Notes:

• Only DMD defines version D_SIMD.
  • Will have to try using the compiles trait
  • At best core.simd will adapt to using SSE or AVX somehow
• No compilers define versions AVX and AVX2 by default.
  • DMD defines both with -mcpu=avx2
  • LDC defines both with -mattr=+avx2