internal/bytealg: vector implementation of equal for riscv64
Provide a vector implementation of equal for riscv64, which is used
when compiled with the rva23u64 profile, or when vector is detected
to be available. Inputs that are 8 byte aligned will still be handled
via a the non-vector code if the length is less than or equal to 64
bytes.
On a Banana Pi F3, with GORISCV64=rva23u64:
│ equal.1 │ equal.2 │
│ sec/op │ sec/op vs base │
Equal/0-8 1.254n ± 0% 1.254n ± 0% ~ (p=1.000 n=10)
Equal/same/1-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.466 n=10)
Equal/same/6-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.689 n=10)
Equal/same/9-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.861 n=10)
Equal/same/15-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.657 n=10)
Equal/same/16-8 21.32n ± 0% 21.33n ± 0% ~ (p=0.075 n=10)
Equal/same/20-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.249 n=10)
Equal/same/32-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.303 n=10)
Equal/same/4K-8 21.32n ± 0% 21.32n ± 0% ~ (p=1.000 n=10)
Equal/same/4M-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.582 n=10)
Equal/same/64M-8 21.32n ± 0% 21.32n ± 0% ~ (p=0.930 n=10)
Equal/1-8 39.16n ± 1% 38.71n ± 0% -1.15% (p=0.000 n=10)
Equal/6-8 51.49n ± 1% 50.40n ± 1% -2.12% (p=0.000 n=10)
Equal/9-8 54.46n ± 1% 53.89n ± 0% -1.04% (p=0.000 n=10)
Equal/15-8 71.81n ± 1% 70.59n ± 0% -1.71% (p=0.000 n=10)
Equal/16-8 69.14n ± 0% 68.21n ± 0% -1.34% (p=0.000 n=10)
Equal/20-8 78.59n ± 0% 77.59n ± 0% -1.26% (p=0.000 n=10)
Equal/32-8 41.55n ± 0% 41.16n ± 0% -0.96% (p=0.000 n=10)
Equal/4K-8 925.5n ± 0% 561.4n ± 1% -39.34% (p=0.000 n=10)
Equal/4M-8 3.110m ± 32% 2.463m ± 16% -20.80% (p=0.000 n=10)
Equal/64M-8 47.34m ± 30% 39.89m ± 16% -15.75% (p=0.004 n=10)
EqualBothUnaligned/64_0-8 32.17n ± 1% 32.11n ± 1% ~ (p=0.184 n=10)
EqualBothUnaligned/64_1-8 79.48n ± 0% 48.24n ± 1% -39.31% (p=0.000 n=10)
EqualBothUnaligned/64_4-8 72.71n ± 0% 48.37n ± 1% -33.48% (p=0.000 n=10)
EqualBothUnaligned/64_7-8 77.12n ± 0% 48.16n ± 1% -37.56% (p=0.000 n=10)
EqualBothUnaligned/4096_0-8 908.4n ± 0% 562.4n ± 2% -38.09% (p=0.000 n=10)
EqualBothUnaligned/4096_1-8 956.6n ± 0% 571.4n ± 3% -40.26% (p=0.000 n=10)
EqualBothUnaligned/4096_4-8 949.6n ± 0% 571.6n ± 3% -39.81% (p=0.000 n=10)
EqualBothUnaligned/4096_7-8 954.2n ± 0% 571.7n ± 3% -40.09% (p=0.000 n=10)
EqualBothUnaligned/4194304_0-8 2.935m ± 29% 2.664m ± 19% ~ (p=0.089 n=10)
EqualBothUnaligned/4194304_1-8 3.341m ± 13% 2.896m ± 34% ~ (p=0.075 n=10)
EqualBothUnaligned/4194304_4-8 3.204m ± 39% 3.352m ± 33% ~ (p=0.796 n=10)
EqualBothUnaligned/4194304_7-8 3.226m ± 30% 2.737m ± 34% -15.16% (p=0.043 n=10)
EqualBothUnaligned/67108864_0-8 49.04m ± 17% 39.94m ± 12% -18.57% (p=0.005 n=10)
EqualBothUnaligned/67108864_1-8 51.96m ± 15% 42.48m ± 15% -18.23% (p=0.015 n=10)
EqualBothUnaligned/67108864_4-8 47.67m ± 17% 37.85m ± 41% -20.61% (p=0.035 n=10)
EqualBothUnaligned/67108864_7-8 53.00m ± 22% 38.76m ± 21% -26.87% (p=0.000 n=10)
CompareBytesEqual-8 51.71n ± 1% 52.00n ± 0% +0.57% (p=0.002 n=10)
geomean 1.469µ 1.265µ -13.93%
│ equal.1 │ equal.2 │
│ B/s │ B/s vs base │
Equal/same/1-8 44.73Mi ± 0% 44.72Mi ± 0% ~ (p=0.426 n=10)
Equal/same/6-8 268.3Mi ± 0% 268.4Mi ± 0% ~ (p=0.753 n=10)
Equal/same/9-8 402.6Mi ± 0% 402.5Mi ± 0% ~ (p=0.209 n=10)
Equal/same/15-8 670.9Mi ± 0% 670.9Mi ± 0% ~ (p=0.724 n=10)
Equal/same/16-8 715.6Mi ± 0% 715.4Mi ± 0% -0.04% (p=0.022 n=10)
Equal/same/20-8 894.6Mi ± 0% 894.5Mi ± 0% ~ (p=0.060 n=10)
Equal/same/32-8 1.398Gi ± 0% 1.398Gi ± 0% ~ (p=0.986 n=10)
Equal/same/4K-8 178.9Gi ± 0% 178.9Gi ± 0% ~ (p=0.853 n=10)
Equal/same/4M-8 178.9Ti ± 0% 178.9Ti ± 0% ~ (p=0.971 n=10)
Equal/same/64M-8 2862.8Ti ± 0% 2862.6Ti ± 0% ~ (p=0.971 n=10)
Equal/1-8 24.35Mi ± 1% 24.63Mi ± 0% +1.16% (p=0.000 n=10)
Equal/6-8 111.1Mi ± 1% 113.5Mi ± 1% +2.17% (p=0.000 n=10)
Equal/9-8 157.6Mi ± 1% 159.3Mi ± 0% +1.05% (p=0.000 n=10)
Equal/15-8 199.2Mi ± 1% 202.7Mi ± 0% +1.74% (p=0.000 n=10)
Equal/16-8 220.7Mi ± 0% 223.7Mi ± 0% +1.36% (p=0.000 n=10)
Equal/20-8 242.7Mi ± 0% 245.8Mi ± 0% +1.27% (p=0.000 n=10)
Equal/32-8 734.3Mi ± 0% 741.6Mi ± 0% +0.98% (p=0.000 n=10)
Equal/4K-8 4.122Gi ± 0% 6.795Gi ± 1% +64.84% (p=0.000 n=10)
Equal/4M-8 1.258Gi ± 24% 1.586Gi ± 14% +26.12% (p=0.000 n=10)
Equal/64M-8 1.320Gi ± 23% 1.567Gi ± 14% +18.69% (p=0.004 n=10)
EqualBothUnaligned/64_0-8 1.853Gi ± 1% 1.856Gi ± 1% ~ (p=0.190 n=10)
EqualBothUnaligned/64_1-8 767.9Mi ± 0% 1265.2Mi ± 1% +64.76% (p=0.000 n=10)
EqualBothUnaligned/64_4-8 839.4Mi ± 0% 1261.9Mi ± 1% +50.33% (p=0.000 n=10)
EqualBothUnaligned/64_7-8 791.4Mi ± 0% 1267.5Mi ± 1% +60.16% (p=0.000 n=10)
EqualBothUnaligned/4096_0-8 4.199Gi ± 0% 6.784Gi ± 2% +61.54% (p=0.000 n=10)
EqualBothUnaligned/4096_1-8 3.988Gi ± 0% 6.676Gi ± 3% +67.40% (p=0.000 n=10)
EqualBothUnaligned/4096_4-8 4.017Gi ± 0% 6.674Gi ± 3% +66.14% (p=0.000 n=10)
EqualBothUnaligned/4096_7-8 3.998Gi ± 0% 6.673Gi ± 3% +66.92% (p=0.000 n=10)
EqualBothUnaligned/4194304_0-8 1.332Gi ± 22% 1.468Gi ± 16% ~ (p=0.089 n=10)
EqualBothUnaligned/4194304_1-8 1.169Gi ± 12% 1.350Gi ± 25% ~ (p=0.075 n=10)
EqualBothUnaligned/4194304_4-8 1.222Gi ± 28% 1.165Gi ± 48% ~ (p=0.796 n=10)
EqualBothUnaligned/4194304_7-8 1.211Gi ± 23% 1.427Gi ± 26% +17.88% (p=0.043 n=10)
EqualBothUnaligned/67108864_0-8 1.274Gi ± 14% 1.567Gi ± 14% +22.97% (p=0.005 n=10)
EqualBothUnaligned/67108864_1-8 1.204Gi ± 14% 1.471Gi ± 13% +22.18% (p=0.015 n=10)
EqualBothUnaligned/67108864_4-8 1.311Gi ± 14% 1.651Gi ± 29% +25.92% (p=0.035 n=10)
EqualBothUnaligned/67108864_7-8 1.179Gi ± 18% 1.612Gi ± 17% +36.73% (p=0.000 n=10)
geomean 1.870Gi 2.190Gi +17.16%
Change-Id: I9c5270bcc6997d020a96d1e97c7e7cfc7ca7fd34
Reviewed-on: https://go-review.googlesource.com/c/go/+/646736
Reviewed-by: Mark Ryan <markdryan@rivosinc.com>
Reviewed-by: Meng Zhuo <mengzhuo1203@gmail.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Mark Freeman <markfreeman@google.com>
Go is an open source programming language that makes it easy to build simple, reliable, and efficient software.
Gopher image by Renee French, licensed under Creative Commons 4.0 Attribution license.
Our canonical Git repository is located at https://go.googlesource.com/go. There is a mirror of the repository at https://github.com/golang/go.
Unless otherwise noted, the Go source files are distributed under the BSD-style license found in the LICENSE file.
Official binary distributions are available at https://go.dev/dl/.
After downloading a binary release, visit https://go.dev/doc/install for installation instructions.
If a binary distribution is not available for your combination of operating system and architecture, visit https://go.dev/doc/install/source for source installation instructions.
Go is the work of thousands of contributors. We appreciate your help!
To contribute, please read the contribution guidelines at https://go.dev/doc/contribute.
Note that the Go project uses the issue tracker for bug reports and proposals only. See https://go.dev/wiki/Questions for a list of places to ask questions about the Go language.