Post

asa
@asap.systems

couldn't read this without constantly thinking about how much nicer it would be with an effect system

vectorware.com@vectorware.com

We are excited to announce that we can successfully use Rust's standard library from the GPU. This has never been done before. www.vectorware.com/blog/rust-st... Supporting Rust's standard library enables existing Rust code to work on the GPU and makes GPU programming feel normal.

vectorware.comRust's standard library on the GPUGPU code can now use Rust's standard library. We share the implementation approach and what this unlocks for GPU programming.

0 reposts14 likes

asa
@asap.systems

the stuff that vectorwave is working on is very cool. however, it feels like there's a mismatch between traditional linear languages like rust (that are eager and sequential, with concurrent and parallel abstractions on top), and the kind of massively parallel code that gpus or simd excels at

Ralith
@ralith.com

The overwhelming majority of code written for those environments is in traditional eager languages. Entirely different paradigms are a cool prospect, but in the mean time I'm looking forward to the massive direct improvement to the status quo that rust-gpu represents.

asa
@asap.systems

yeah i agree. lazy execution-independent languages are also mismatched performance-wise with cpus designed to execute c. rust is a systems language that effectively compromises with "you can't replace all of modern computing infrastructure," and linear eager execution is part of that

asa
@asap.systems

intuitively, languages like haskell and bend/hvm seem more suited to general purpose computing on top of both cpus and gpus *i'm extremely far from an expert on this