Hi. I'm working in the OpenBSD/Sparc64 port of Racket. Hopefully, there
is only one remaining bug related to foreign/ffi. It was pretty easy.
I ran racket-benchmarks/shootout and it was amazingly slow, I gave up
after the first test and I cancelled the script. Racket's GMP code lacks
the Sparc64 assembler code, so I added it. I tried even to update the
GMP code during various days and it worked fine on
amd64/sparc64/linux-ppc64 but broke linux-armv7. Anyway, even with the
assembler code, the benchmarks were slow as hell. I guess that the
critical part of the good Racket performance on amd64/i386 is mostly the
JIT.
Don't take my words as criticism, I'm quite happy with Racket. It makes
easy my work as maintainer. It's an amazing project from a small team of
developers.
So, rants apart, my question is: there is a long-term plan to remove the
GMP/Lightning bundled code and to use directly the vanilla code from the
external projects?. That would improve the performance of Racket on
non-mainstream platforms and distros with non-bundled-libraries policy
(like Fedora/CentOS) could include Racket again. Lightning supports a
bunch of architectures.
--
You received this message because you are subscribed to the Google Groups "Racket
Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/racket-dev/mqlm7m%24qe1%241%40ger.gmane.org.
For more options, visit https://groups.google.com/d/optout.
