Language abstractions and scheduling techniques for efficient execution of parallel algorithms on multicore hardware

Multicore hardware has been mainstream for more than a decade. So I am not sure that “week” is the right timescale here :slight_smile:

No, beyond trivial examples, it won’t work. I think that you really need to capture the current continuation.

I understand that it can be tricky to implement for many reasons, but I don’t know of another way to expose parallel constructs that you can smoothly introduce in the middle of conventional sequential code, i.e. without requiring to manually CPS-transform the entire program.