upvote
I'd be interested to see how this would pan out.

Seems to describe something analogous instruction level parallelism, which traditional CPUs made this millennium have as well via instruction pipelines, branch prediction, and speculative execution. Some of which turned out to not work too well so now we generally run hobbled CPUs with Spectre/Meltdown mitigation enabled.

The reason why we tend to reach for difficult abstractions for core level parallelism is that most automatic/safe approaches tend to stumble pretty hard on Amdahl's law, where most parallel code is consequently almost entirely designed around minimizing critical sections.

I don't fully see how inets would resolve Amdahl's law.

reply
I think you're thinking about this the wrong way. This isnt about surpassing theoretical parallelism limits. Its about gaining parallelism without having to be explicit about it. If you look at the solve function from the page, its doing a brute force search of the 5 queens problem. In the naieve implementation, this is done serially. But with the different approach to compiling the implementation, we can essentially "automatically parallelize" it. Thats the difference between the animation on the left and right. The thing is that in order to handle fine grained parallelism like this you need better/different hardware primitives so that you can split computation at a more fine grain, which is why they needed a new hardware model/isa. This is how they achieved the 4-5x speed up: by automatically parallizing a serially implemented function. If you were to compare a parallelized function to their parallelized function, I would expect to see a smaller speedup. Or at least as I understand it. Correct me if I'm wrong.
reply
I read through the first half, but the graph operations are really confusing me. What's the difference between a principal port and auxiliary ports? A lot of the rewrites are swapping them around so I'm confused what their significance is.
reply
This is the key feature of the architecture. Computation is shown to be localizable and independent in terms of 'active pairs' which are defined as elements connected via their 'principle port'. Since the design permits only one 'principle port' per element, all possible graphs that we can arrive at with this formalism can only have components that 'exclusively' participate in a single 'active pair'.

So there is 'no sharing of state' baked into the computational model by design, enforced by this very simple and elegant design choice.

reply
deleted
reply
I was a bit surprised by Vine's syntax. I expected something a lot more like Lisp.
reply