Parallelism with array languages

First of a series in which
I look at how hardware parallelism evolved.
I will argue for the conceptual advantages of whole array programming.
I will explain the constructs and idioms of such programming.
I will give examples of the performance gains to be had.

[@theskoomacat7849]

Thank you for this lecture! I enjoy your videos

[@slightlygruff]

pascal is so intuitive. It'll be back

[@_jamesdphillips]

Great lecture, thank you!

[@ryandepp7640]

Could you make a video on the potential of economic planning in developing countries, perhaps mentioning the Feldman model?

[@PC42190]

Hey Paul, do you find accurate or realistic Henryk Grossman’s breakdown theory?

[@eleanorjones8633]

Grossman's math is predicated on the assumption of a 100% profit reinvestment rate which is grossly inaccurate even during boom periods so it's null.
His argument is essentially that if all profit is reinvestment then eventually the capital stock will grow to the point where the requirements of replacing depreciation leave nothing leftover for the capitalists themselves which is mathematically true but also besides the point since that doesn't ever happen in real capitalist economies.

[@eleanorjones8633]

Don't know why my past comment vanished so I'm writing it again.
Grossman's math in arguing for the inevitability of capitalist breakdown is predicated on the assumption of a 100% profit reinvestment rate. Empirical data shows this is a gross overestimation even in boom periods, so the model is null.
He claims that if all profit is reinvested eventually the capital stock will grow to the point that the requirements of replacing depreciation leave nothing left for the capitalists themselves. This is mathematically true but politically irrelevant since this never actually happens in real capitalist economies.
Ironically this analytical error of just taking it as given that capitalists will always accumulate as much as possible was committed by Marx himself, and was one of his biggest mistakes.

[@aaronblain6377]

So, are we able to avoid the boundary check because it can now be done in hardware, but C doesn't have a way to produce the relevant opcode?
You mentioned functional programming. How might program design need to evolve more generally as we get further away from the single-threaded procedural paradigm? Are the costs of parallelism essentially solvable or will they continue to mount up?

[@ArilandoArilando]

What about Rust?

[@mek101whatif7]

🦀🦀🦀🦀🦀🦀🦀🦀🦀🦀🦀

[@kasunkt]

Please do a video on AI and Marxism

[@anotherinternetperson7410]

Mr cockshott can you make a video on Marx’s saying “from each according to his abilities to each according to his needs”

[@mecha37000fighter]

Love it🫡