internal/bytealg: vector implementation of indexbyte for riscv64

Provide a vector implementation of indexbyte for riscv64, which is used
when compiled with the rva23u64 profile, or when vector is detected
to be available. Inputs that are smaller than 24 bytes will continue
to use the non-vector path.

On a Banana Pi F3, with GORISCV64=rva23u64:

                │  indexbyte.1  │             indexbyte.2              │
                │    sec/op     │    sec/op     vs base                │
IndexByte/10-8     52.68n ±  0%   47.26n ±  0%  -10.30% (p=0.000 n=10)
IndexByte/32-8     68.62n ±  0%   47.02n ±  0%  -31.49% (p=0.000 n=10)
IndexByte/4K-8    2217.0n ±  0%   420.4n ±  0%  -81.04% (p=0.000 n=10)
IndexByte/4M-8    2624.4µ ±  0%   767.5µ ±  0%  -70.75% (p=0.000 n=10)
IndexByte/64M-8    68.08m ± 10%   47.84m ± 45%  -29.73% (p=0.004 n=10)
geomean            17.03µ         8.073µ        -52.59%

                │ indexbyte.1  │               indexbyte.2               │
                │     B/s      │      B/s        vs base                 │
IndexByte/10-8    181.0Mi ± 0%    201.8Mi ±  0%   +11.48% (p=0.000 n=10)
IndexByte/32-8    444.7Mi ± 0%    649.1Mi ±  0%   +45.97% (p=0.000 n=10)
IndexByte/4K-8    1.721Gi ± 0%    9.076Gi ±  0%  +427.51% (p=0.000 n=10)
IndexByte/4M-8    1.488Gi ± 0%    5.089Gi ±  0%  +241.93% (p=0.000 n=10)
IndexByte/64M-8   940.3Mi ± 9%   1337.8Mi ± 31%   +42.27% (p=0.004 n=10)
geomean           727.1Mi         1.498Gi        +110.94%

Change-Id: If7b0dbef38d76fa7a2021e4ecaed668a1d4b9783
Reviewed-on: https://go-review.googlesource.com/c/go/+/648856
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
Reviewed-by: Mark Freeman <markfreeman@google.com>
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

0 files changed, 0 insertions, 0 deletions