Oh the shame!!!!
Thank you for pointing that out :)
On Mon, 22 Jun 2020, 20:03 Martin Schnabel, <m...@mb0.org> wrote:
> Let me quickly point out that your logic is inverted in the second code
> snippet because you break on the condition. That is most likely the
> reason for the difference. I guess the second example exits on the first
> iteration.
>
> for pos := 0; true; pos++ {
> if pos < suffixLen && pos < previousSuffixLen {
> continue
> }
> break
> }
>
> On 22.06.20 14:15, Yonatan Ben-Nes wrote:
> > Hi,
> >
> > I'm encountering a weird issue which I fail to explain. It boils down to
> > 2 almost identical functions which give wildly different benchmark
> results.
> >
> > The difference between the functions is that at the slow func there is a
> > for loop like this:
> > for pos := 0; pos < suffixLen && pos < previousSuffixLen; pos++ {
> > }
> >
> > While at the fast func the loop is like this:
> > for pos := 0; true; pos++ {
> > if pos < suffixLen && pos < previousSuffixLen {
> > break
> > }
> > }
> > * do note that the check at both versions is the same.
> >
> > Other than that the functions are identical but when I benchmark them I
> get:
> > $ go test -bench=Weird .
> > goos: linux
> > goarch: amd64
> > BenchmarkWeirdFast-4 280394 3602 ns/op
> > BenchmarkWeirdSlow-4 1866 618953 ns/op
> >
> > It's extra weird since getting the condition check into the loop should
> > be the slower form, or at least that what I intuitively thought.
> >
> > Also, the extra time cost is divided between the loop section and an
> > append function call which run immediately at the start of the function
> > and is not connected in any way to the for loop.
> > And it's extra weird as the append cost also get smaller if instead of
> > reading the test text from a file I just declare a string variable with
> > the same text for the benchmark (mind you that I b.ResetTimer() after
> > doing either of those options).
> >
> > I attached the files to this message as they will make everything more
> > clear and I also added comments describing what happen at the right
> points.
> >
> > I'm using Ubuntu 18.04 and go is version 1.14.2, and I also checked it
> > on a GCP compute engine and got the same results.
> >
> > Any thoughts will be much appreciated as I'm totally confused here! :)
> >
> > --
> > 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
> > <mailto:golang-nuts+unsubscr...@googlegroups.com>.
> > To view this discussion on the web visit
> >
> https://groups.google.com/d/msgid/golang-nuts/3cf40b00-96ec-42a6-8093-7f1c27bfd93fn%40googlegroups.com
> > <
> https://groups.google.com/d/msgid/golang-nuts/3cf40b00-96ec-42a6-8093-7f1c27bfd93fn%40googlegroups.com?utm_medium=email&utm_source=footer
> >.
>
--
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/CALtqQdvudO1oHKJCfRt_0YXv8nEp7jhZf671HijVqNFFOfcLbQ%40mail.gmail.com.
