When lowering some ops on amd64 we generate additional NEGQ. This may result in code like this: NEGQ R12 NEGQ R12 Optimize it away. Gain is not significant, about ~0.5% gain in geomean in compress/flate and 200 bytes codesize reduction in go tool. Full results below: name old time/op new time/op delta Encode/Digits/Huffman/1e4-6 65.8µs ± 0% 65.7µs ± 0% -0.21% (p=0.010 n=10+9) Encode/Digits/Huffman/1e5-6 633µs ± 0% 632µs ± 0% ~ (p=0.370 n=8+9) Encode/Digits/Huffman/1e6-6 6.30ms ± 1% 6.29ms ± 1% ~ (p=0.796 n=10+10) Encode/Digits/Speed/1e4-6 281µs ± 0% 280µs ± 1% -0.34% (p=0.043 n=8+10) Encode/Digits/Speed/1e5-6 2.66ms ± 0% 2.66ms ± 0% -0.09% (p=0.043 n=10+10) Encode/Digits/Speed/1e6-6 26.3ms ± 0% 26.3ms ± 0% ~ (p=0.190 n=10+10) Encode/Digits/Default/1e4-6 554µs ± 0% 557µs ± 0% +0.46% (p=0.001 n=9+10) Encode/Digits/Default/1e5-6 8.63ms ± 1% 8.62ms ± 1% ~ (p=0.912 n=10+10) Encode/Digits/Default/1e6-6 92.7ms ± 1% 92.2ms ± 1% ~ (p=0.052 n=10+10) Encode/Digits/Compression/1e4-6 558µs ± 1% 557µs ± 1% ~ (p=0.481 n=10+10) Encode/Digits/Compression/1e5-6 8.58ms ± 0% 8.61ms ± 1% ~ (p=0.315 n=8+10) Encode/Digits/Compression/1e6-6 92.3ms ± 1% 92.4ms ± 1% ~ (p=0.971 n=10+10) Encode/Twain/Huffman/1e4-6 89.5µs ± 0% 89.0µs ± 1% -0.48% (p=0.001 n=9+9) Encode/Twain/Huffman/1e5-6 727µs ± 1% 728µs ± 0% ~ (p=0.604 n=10+9) Encode/Twain/Huffman/1e6-6 7.21ms ± 0% 7.19ms ± 1% ~ (p=0.696 n=8+10) Encode/Twain/Speed/1e4-6 320µs ± 1% 321µs ± 1% ~ (p=0.353 n=10+10) Encode/Twain/Speed/1e5-6 2.63ms ± 0% 2.62ms ± 1% -0.33% (p=0.016 n=8+10) Encode/Twain/Speed/1e6-6 25.8ms ± 0% 25.8ms ± 0% ~ (p=0.360 n=10+8) Encode/Twain/Default/1e4-6 677µs ± 1% 671µs ± 1% -0.88% (p=0.000 n=10+10) Encode/Twain/Default/1e5-6 10.5ms ± 1% 10.3ms ± 0% -2.06% (p=0.000 n=10+10) Encode/Twain/Default/1e6-6 113ms ± 1% 111ms ± 1% -1.96% (p=0.000 n=10+9) Encode/Twain/Compression/1e4-6 688µs ± 0% 679µs ± 1% -1.30% (p=0.000 n=7+10) Encode/Twain/Compression/1e5-6 11.6ms ± 1% 11.3ms ± 1% -2.10% (p=0.000 n=10+10) Encode/Twain/Compression/1e6-6 126ms ± 1% 124ms ± 0% -1.57% (p=0.000 n=10+10) [Geo mean] 3.45ms 3.44ms -0.46% name old speed new speed delta Encode/Digits/Huffman/1e4-6 152MB/s ± 0% 152MB/s ± 0% +0.21% (p=0.009 n=10+9) Encode/Digits/Huffman/1e5-6 158MB/s ± 0% 158MB/s ± 0% ~ (p=0.336 n=8+9) Encode/Digits/Huffman/1e6-6 159MB/s ± 1% 159MB/s ± 1% ~ (p=0.781 n=10+10) Encode/Digits/Speed/1e4-6 35.6MB/s ± 0% 35.7MB/s ± 1% +0.34% (p=0.020 n=8+10) Encode/Digits/Speed/1e5-6 37.6MB/s ± 0% 37.7MB/s ± 0% +0.09% (p=0.049 n=10+10) Encode/Digits/Speed/1e6-6 38.0MB/s ± 0% 38.0MB/s ± 0% ~ (p=0.146 n=10+10) Encode/Digits/Default/1e4-6 18.0MB/s ± 0% 18.0MB/s ± 0% -0.45% (p=0.002 n=9+10) Encode/Digits/Default/1e5-6 11.6MB/s ± 1% 11.6MB/s ± 1% ~ (p=0.644 n=10+10) Encode/Digits/Default/1e6-6 10.8MB/s ± 1% 10.8MB/s ± 1% +0.51% (p=0.044 n=10+10) Encode/Digits/Compression/1e4-6 17.9MB/s ± 1% 17.9MB/s ± 1% ~ (p=0.468 n=10+10) Encode/Digits/Compression/1e5-6 11.7MB/s ± 0% 11.6MB/s ± 1% ~ (p=0.322 n=8+10) Encode/Digits/Compression/1e6-6 10.8MB/s ± 1% 10.8MB/s ± 1% ~ (p=0.983 n=10+10) Encode/Twain/Huffman/1e4-6 112MB/s ± 0% 112MB/s ± 1% +0.42% (p=0.002 n=8+9) Encode/Twain/Huffman/1e5-6 138MB/s ± 1% 137MB/s ± 0% ~ (p=0.616 n=10+9) Encode/Twain/Huffman/1e6-6 139MB/s ± 0% 139MB/s ± 1% ~ (p=0.652 n=8+10) Encode/Twain/Speed/1e4-6 31.3MB/s ± 1% 31.2MB/s ± 1% ~ (p=0.342 n=10+10) Encode/Twain/Speed/1e5-6 38.0MB/s ± 0% 38.1MB/s ± 1% +0.33% (p=0.011 n=8+10) Encode/Twain/Speed/1e6-6 38.8MB/s ± 0% 38.7MB/s ± 0% ~ (p=0.325 n=10+8) Encode/Twain/Default/1e4-6 14.8MB/s ± 1% 14.9MB/s ± 1% +0.88% (p=0.000 n=10+10) Encode/Twain/Default/1e5-6 9.48MB/s ± 1% 9.68MB/s ± 0% +2.11% (p=0.000 n=10+10) Encode/Twain/Default/1e6-6 8.86MB/s ± 1% 9.03MB/s ± 1% +1.97% (p=0.000 n=10+9) Encode/Twain/Compression/1e4-6 14.5MB/s ± 0% 14.7MB/s ± 1% +1.31% (p=0.000 n=7+10) Encode/Twain/Compression/1e5-6 8.63MB/s ± 1% 8.82MB/s ± 1% +2.17% (p=0.000 n=10+10) Encode/Twain/Compression/1e6-6 7.92MB/s ± 1% 8.05MB/s ± 1% +1.59% (p=0.000 n=10+10) [Geo mean] 29.0MB/s 29.1MB/s +0.47% // symSizeComp `which go` go_old: section differences: global text (code) = 203 bytes (0.005131%) read-only data = 1 bytes (0.000057%) Total difference 204 bytes (0.003297%) Change-Id: Ie2cdfa1216472d78694fff44d215b3b8e71cf7bf Reviewed-on: https://go-review.googlesource.com/102277 Run-TryBot: Ilya Tocar <ilya.tocar@intel.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org> |
||
|---|---|---|
| .github | ||
| api | ||
| doc | ||
| lib/time | ||
| misc | ||
| src | ||
| test | ||
| .gitattributes | ||
| .gitignore | ||
| AUTHORS | ||
| CONTRIBUTING.md | ||
| CONTRIBUTORS | ||
| LICENSE | ||
| PATENTS | ||
| README.md | ||
| favicon.ico | ||
| robots.txt | ||
README.md
The Go Programming Language
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 3.0 Attributions 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.
Download and Install
Binary Distributions
Official binary distributions are available at https://golang.org/dl/.
After downloading a binary release, visit https://golang.org/doc/install or load doc/install.html in your web browser for installation instructions.
Install From Source
If a binary distribution is not available for your combination of operating system and architecture, visit https://golang.org/doc/install/source or load doc/install-source.html in your web browser for source installation instructions.
Contributing
Go is the work of hundreds of contributors. We appreciate your help!
To contribute, please read the contribution guidelines: https://golang.org/doc/contribute.html
Note that the Go project uses the issue tracker for bug reports and proposals only. See https://golang.org/wiki/Questions for a list of places to ask questions about the Go language.