On Wed, Feb 5, 2020 at 8:43 PM Carlos Amedee <car...@golang.org> wrote:
> We have just released go1.14rc1, a release candidate version of Go 1.14.
I am glad to report substantial improvements for a set of benchmarks of the
gocc
compiler[0], which is an experimental/WIP C compiler implemented in Go that
can
also produce Go code.
Note: The benchmarks are not benchmarking the compiler but the code
produced by the gocc compiler.
----
gocc commit bf916e5600, Tue Dec 24 17:01:06 2019 +0100
Go 1.13.7 linux/amd64
...
all_test.go:2092: goccgo perlin.c 411.430ms true true
true 13.769
...
all_test.go:2099: Considered tests: 24/24
all_test.go:2107: gcc 1.926479916s 1.000
all_test.go:2107: gocc 2.084863963s 1.082
all_test.go:2107: goccgo 6.272091508s 3.256
Go 1.14rc1 linux/amd64
TestCompCert: all_test.go:2092: goccgo perlin.c 159.321ms
true true true 5.396
...
TestCompCert: all_test.go:2099: Considered tests: 24/24
TestCompCert: all_test.go:2107: gcc 1.945554301s 1.000
TestCompCert: all_test.go:2107: gocc 2.073950104s 1.066
TestCompCert: all_test.go:2107: goccgo 4.853355485s 2.495
----
The cumulative index (gcc = 1) of executing a set of 24 benchmarks for the
Go backend went from 3.256
to 2.495 (-23%).
The worst performing test, perlin.c, was executing 13.769 times slower than
the code produced
by gcc with Go 1.13.7 and the Go backend. With Go 1.14rc1 this went down to
5.396 time slower (-60%).
Full results are enclosed in the attachments.
----
[0]: https://godoc.org/modernc.org/gocc
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/golang-nuts/CAA40n-UcH2n96EL4PU7%2B-Qbd6TcghZ%3DG0r2H6b4a2Q13E0%3DdTQ%40mail.gmail.com.
log-go1.13.7
Description: Binary data
log-go1.14rc1
Description: Binary data
