Hi Johannes, see also this wiki page: https://ghc.haskell.org/trac/ghc/wiki/Performance/Runtime. It mentions the `gcd` regression that you found (and has some analysis), but not the others (`tak` and `exp3_8`). Maybe you could start with updating that page. Thomas On Thu, Jan 7, 2016 at 9:55 AM, Simon Peyton Jones <simonpj@microsoft.com> wrote:
Yes we’d love help with investigating, characterising, and fixing performance bugs.
Simon
*From:* Haskell-Cafe [mailto:haskell-cafe-bounces@haskell.org] *On Behalf Of *Thomas Miedema *Sent:* 07 January 2016 00:52 *To:* Johannes Waldmann <johannes.waldmann@htwk-leipzig.de> *Cc:* Joachim Breitner <mail@joachim-breitner.de>; Haskell cafe < haskell-cafe@haskell.org> *Subject:* Re: [Haskell-cafe] long-range performance comparisons (GHC over the years)
I recently noticed a performance problem
...
What worries me is that such a regression had been sitting there for over a year
There are 215 open *runtime performance <https://ghc.haskell.org/trac/ghc/query?status=!closed&failure=Runtime+performance+bug&order=id>* tickets (out of a total of 1655 open tickets, that makes 13%).
Compared to say new typesystem features, they don't get much attention. Only a few were fixed <https://ghc.haskell.org/trac/ghc/query?failure=Runtime+performance+bug&resolution=fixed&milestone=8.0.1&milestone=7.10.3&milestone=7.10.2> this year, most of them by Joachim Breitner (the new performance tsar <https://ghc.haskell.org/trac/ghc/wiki/CodeOwners#CurrentOwners>?). He also created https://perf.haskell.org/ghc <https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fperf.haskell.org%2fghc%2f&data=01%7c01%7csimonpj%40064d.mgd.microsoft.com%7cc3af688d2c234b3c19b608d316fcb626%7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=Ik7nCUHxNalCIxrv9E8dq2od9ed%2bbbLXJ%2fJJyHLCIhQ%3d> .
In case you want to help out, start here: https://ghc.haskell.org/trac/ghc/wiki/Newcomers.