Collect benchmark numbers for 0.1

[brson]

We need to know how our numbers now compare to those from last summer and those from gcc.

[kud1ing]

Are these numbers available somewhere?

[brson]

Here is what I collected, 32-bit target, optimizations on:

https://gist.github.com/1604055

oldrust is a version from may 2010.

[brson]

*May 2011

[kud1ing]

Oh ... interesting, thanks. Do we have an understanding yet what caused those regressions?

See also my Java, GHC numbers for the N-Body benchmark:
https://mail.mozilla.org/pipermail/rust-dev/2011-December/001159.html

In general i find the comparison with GHC interesting, because to me that's one of Rust's advantages.
You can do pure-functional programming, but you are not locked into it when it does not fit the problem.

[brson]

I haven't investigated yet - maybe this weekend. I saw your benchmark numbers. Thanks for making them.

[brson]

Hopefully we can squeeze out some performance improvements before 0.1

[brson]

shootout-binarytrees has terrible performance because of #1493. I updated that benchmark to avoid shared boxes.

[brson]

See also #1527

[brson]

@Lenny222: I updated the benchmarks to account for the two issues above and took new numbers: https://gist.github.com/1614498

[kud1ing]

That's sort of promising, thanks.

[ghost]

I would like to run again on my imac on latest rust, I have the rust code in src/test/bench can you please provide the corresponding C code, if you still have for gcc.

[brson]

@webconv The shootout source for other languages is here: https://alioth.debian.org/snapshots.php?group_id=30402