I'm a huge fan of Feynman and his autobiographies - some of my favorite books. There are a number of YouTube videos that purport to be Feynman's audio lectures, but they're actually AI generated readings of his script. Against my better instincts I listened to a couple and it was surreal experience.
On one hand, they were actually pretty decent simulacra and I loved hearing Feynman's voice and mannerisms; but on the other hand, they... weren't... real. What am I even listening to?
I agree that something like *lisp would be a welcome relief from the hairball that is CUDA. while the lisp part was actually compiled, the parallel part reduced pretty 1:1 onto Paris, (the 'assembly language' designed by Steele). that was was basically a jump table into the microcode, which was just a big control word issued by (what I remember vaguely was being an amd 2900 bit slice) sequencer.
today I guess you'd translate those vector instruction into MLIR and hope for some layout and fusion.
but absolutely, you can just use tensors like normal variables. what a awful regression
I'm a huge fan of Feynman and his autobiographies - some of my favorite books. There are a number of YouTube videos that purport to be Feynman's audio lectures, but they're actually AI generated readings of his script. Against my better instincts I listened to a couple and it was surreal experience.
On one hand, they were actually pretty decent simulacra and I loved hearing Feynman's voice and mannerisms; but on the other hand, they... weren't... real. What am I even listening to?
We are in for a weird future.
Angela Collier's video about Feynman completely changed my view of him.
https://news.ycombinator.com/item?id=48915709 one of the Connection Machines is in Jurassic Park. Really neat article!
Danny Hillis was an MIT colleague of a friend of mine. Didn't know until now that Connection Machine had Darpa funding.
Star Lisp would be great for the modern heterogeneous computing landscape, and being compiled.
Can you explain this as Richard would have?
I agree that something like *lisp would be a welcome relief from the hairball that is CUDA. while the lisp part was actually compiled, the parallel part reduced pretty 1:1 onto Paris, (the 'assembly language' designed by Steele). that was was basically a jump table into the microcode, which was just a big control word issued by (what I remember vaguely was being an amd 2900 bit slice) sequencer.
today I guess you'd translate those vector instruction into MLIR and hope for some layout and fusion.
but absolutely, you can just use tensors like normal variables. what a awful regression
banging on the door like a madman where's the clock? where's the clock we all donated money for? tell us about the clock you cowards!
Yeah.... that was a cool idea. Too bad it'll never be built.
(Supposedly they're building it in West Texas - last update in 2024)
Should be here about now-ish.
Written in 2017, but always a good read.