Higher-order Virtual Machine (HVM). New functional programming VM based on Interaction Nets.
anon 0x137 said in #1240 14mo ago:
Not to turn this place into Lambda the Ultimate, but if you're into lambda calculus autism and optimal reduction order stuff, this seems very cool. Apparently new techniques based on Interaction Networks allow a lazy non garbage collected vm to reduce some statements exponentially faster, and massively parallel. We might see functional programming finally realizing it's raison d'etre of taking much more full advantage of parallel computing ("liberating programming from von Neumann style").
I remember some years ago reading a paper about using advanced forms of lazy evaluation to collapse an arbitrary tower of interpreters interpreting their own source code into at most a constant factor slowdown. This VM is showing similar exponential speedups vs things like GHC for certain problems, and is much more aimed at production than those old academic prototypes were. This might sound really obscure, but collapsing towers of interpreters has big implications for expressive and performant programming with deeper abstractions.
Even so, going the other direction into less abstraction and more hands-on knowledge of how you're moving bits around is also very undervalued.
I remember some years ago reading a paper about using advanced forms of lazy evaluation to collapse an arbitrary tower of interpreters interpreting their own source code into at most a constant factor slowdown. This VM is showing similar exponential speedups vs things like GHC for certain problems, and is much more aimed at production than those old academic prototypes were. This might sound really obscure, but collapsing towers of interpreters has big implications for expressive and performant programming with deeper abstractions.
Even so, going the other direction into less abstraction and more hands-on knowledge of how you're moving bits around is also very undervalued.
Not to turn this pla
anon 0x138 said in #1242 14mo ago:
As an old lambda calculus and Haskell guy, I think this is very cool. It would take a ton of effort to realize its potential (as with all PL stacks), but I'll certainly be watching it. It would be incredible to see ASICs optimized for Interaction Networks to realize maximum parallelism on chip.
As an old lambda cal
anon 0x29f said in #1794 12mo ago:
Oh interesting, these appear to be the same guys behind the Bend language that has been doing the rounds recently:
https://github.com/HigherOrderCO/Bend
(I have heard Bend described as "the syntax of Python except it's run by raising matrices to powers on GPUs.")
https://github.com/HigherOrderCO/Bend
(I have heard Bend described as "the syntax of Python except it's run by raising matrices to powers on GPUs.")
referenced by: >>1797
Oh interesting, thes
anon 0x137 said in #1797 12mo ago:
>>1794
Yeah Bend is the language built on the HVM. I haven't looked into it yet but I'd love to try it out when I get a chance. The other one I've recently been wanting to check out is Zig. It really does look like C with the problems fixed. But I have no great project ideas for either of these languages. My best ideas are web application software and AI research code. That means Go and Python, which are both a bit meh in their own ways.
Yeah Bend is the language built on the HVM. I haven't looked into it yet but I'd love to try it out when I get a chance. The other one I've recently been wanting to check out is Zig. It really does look like C with the problems fixed. But I have no great project ideas for either of these languages. My best ideas are web application software and AI research code. That means Go and Python, which are both a bit meh in their own ways.
Yeah Bend is the lan