On Wed, Jul 13, 2016 at 5:06 PM, Eliot Miranda <[hidden email]> wrote:
Well, I won't stand in the way of you gaining more performance; quite the opposite; I encourage it. If you have energy here please apply it. But it would be nice if you added the speedups to our code base rather than introduced a dependency on GMP.
Yes, it would be awesome to improve large integer performance without introducing a dependency on GMP. I think the smallest thing that would have the biggest impact in performance is implementing highly optimized multiplication (and perhaps division) for x86/x86_64/arm. If it was not because of the license problems, we could just copy a couple of .S files straight from GMP source code...
Now I'm very much focused on the Smalltalk side of my project. But if no one does it first, I'll give it a shot.
I was going to suggest working with Sista, but that leaves the interpreter VMs out in the cold.
Yeah, I'm looking forward to Sista. I was planning to implement some primitives for what I think are the main bottlenecks (mostly operations with arrays of integers or arrays of integers mod n where n fits in a 32 bits register for example), but I'm kind of (speculatively) hoping that speculative optimizations might achieve similar performance as writing those primitives in C \o/
Locked
