Hacker News Reader

It's interesting how "array programming" (I think) was a sort of prediction of what we would be doing with GPUs today. (At least that's what occurred to me just now.)

Does anyone with this expertise have a sense for how APL relates to modern GPU programming? Is there an influence? Or is it a completely different model?

View on HN

Array programming seems adaptable to any form of parallel processing: SIMD instructions, multi-core, or GPUs. I think it's because, in addition to their inherent parallelism, primitives are so simple that they make very few demands on an implementation. GPU code can be a lot more flexible than array programming because the processors have a bit of independence, so array primitives are not necessarily the best way to use a GPU, but they're pretty good and the limitations might make it easier for a human to think about.

Reply on news.ycombinator.com