by nicolas.riesco:

Before filing a bug, please check whether it has been fixed since the
latest release. Search the issue tracker and check that you're running the
latest version of Go:

Run "go version" and compare against
http://golang.org/doc/devel/release.html  If a newer version of Go exists,
install it and retry what you did to reproduce the problem.

Thanks.

What does 'go version' print?
go version devel +08dcdcdb757b Thu Mar 13 14:04:29 2014 -0700 linux/amd64

What steps reproduce the problem?
I'm attaching a file with benchmarks to illustrate how the optimization can be achieved.
In few words, I'm proposing that loops as this one:

    for j := range s {
        s[j].X, s[j].Y, s[j].Z = 1, 2, 3
    }

could be optimised by the compiler and implemented as:

    for j := range s {
        p := &s[j]
        p.X, p.Y, p.Z = 1, 2, 3
    }

What happened?

The unoptimised benchmarks:

    BenchmarkLoop       10000        182743 ns/op
    BenchmarkLoopLen       10000        170704 ns/op
    BenchmarkLoopRange       10000        170395 ns/op

run slower than the optimised ones:

    BenchmarkLoopPointer       10000        136950 ns/op
    BenchmarkLoopPointerRange       10000        111240 ns/op      

What should have happened instead?



Please provide any additional information below.

This optimization is more important that removing bound checks. I've run the benchmarks
with and without bound checks and the effect of this optimization is still important:

$ go test -bench=.  loops_test.go
testing: warning: no tests to run
PASS
BenchmarkLoop       10000        182743 ns/op
BenchmarkLoopPointer       10000        136950 ns/op
BenchmarkLoopLen       10000        170704 ns/op
BenchmarkLoopRange       10000        170395 ns/op
BenchmarkLoopPointerRange       10000        111240 ns/op
ok      command-line-arguments    7.802s

$ go test -bench=. -gcflags=-B loops_test.go
testing: warning: no tests to run
PASS
BenchmarkLoop       10000        151952 ns/op
BenchmarkLoopPointer       10000        123730 ns/op
BenchmarkLoopLen       10000        134355 ns/op
BenchmarkLoopRange       10000        134049 ns/op
BenchmarkLoopPointerRange       10000        110534 ns/op
ok      command-line-arguments    6.620s

Attachments:

1. loops_test.go (4393 bytes)