[CM] Performance of S7

David O'Toole deeteeoh1138 at gmail.com
Wed Dec 30 04:24:19 PST 2020


I'm not sure this is 100% relevant, but: I noticed a big speed improvement
when I compiled without GUI support (even though I had been suppressing the
GUI with command line -b option.)

On Wed, Dec 30, 2020 at 7:14 AM Massimiliano Gubinelli <
m.gubinelli at gmail.com> wrote:

> Dear all,
>
>  I wanted to ask about the performance of S7 and how to understand
> possible causes of slowness. My use case is the following. I wanted to
> experiment in replacing the current scheme interpreter in GNU TeXmacs (
> www.texmacs.org) with S7. Currently we are using Guile 1.8. The program
> is an interactive editor for structured documents where a large part of the
> GUI code is written in scheme (think of a graphical Emacs). I've managed to
> have an (almost) working version of TeXmacs with S7 where I can work a bit
> with the editor. Not all the functionalities are working yet but enough to
> make some speed comparision and I noticed that S7 is much slower than Guile
> 1.8 for our use-case. I didn't performed any hard timing but my sensation
> is that slowdown is 2-3x. This surprised me because in other very shallow
> tests, like running (fib 40) S7 is ~8x faster than Guile.
>
> What I noted is that S7 is slower than Guile 1.8 in loading code (we have
> ~100.000 lines of scheme to load at startup) and also in running recursive
> algorithms on trees which require hash-table lookups on symbols (we have a
> small compiler for dynamic menus which has to run quite fast at every
> keypress).
>
> * Do you have any suggestions on which is going on and if there are
> directions where to look for speed improvements?
>
> * Is this relative performance with Guile 1.8 on tree/symbol related tasks
> to be expected?
>
> * Do somebody has some experience in comparing performances of S7 with
> other scheme interpreters in real-world use-cases?
>
> S7 is a very cool project. I was really hoping to be able to use it in
> TeXmacs.
>
> Best regards,
> Massimiliano Gubinelli
>
>
> _______________________________________________
> Cmdist mailing list
> Cmdist at ccrma.stanford.edu
> https://cm-mail.stanford.edu/mailman/listinfo/cmdist
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://cm-mail.stanford.edu/pipermail/cmdist/attachments/20201230/35ac515b/attachment.html>


More information about the Cmdist mailing list