Writing geophysical | military signal and image processing applications on custom DSP clusters is suprisingly straightforward and doesn't need C++.
It's a RISC architecture optimised for DSP | FFT | Array processing with the basic simplification that char text is for hosts, integers and floats are at least 32 bit and 32 bits (or 64) is the smallest addressable unit.
Fantastic architecture to work with for numerics, deep computational pipelines, once "primed" you push in raw aquisition samples in chunks every clock cycle and extract processed moving window data chunks every clock cycle.
A single ASM instruction in a cycle can accumulate totals from vector multiplication and modulo update indexes on three vectors (two inputs and and out).
Not your mama's brainfuck.
s/_BigInt/_BitInt/
https://thephd.dev/conformance-should-mean-something-fputc-a...
Me? I just dabble with documenting an unimplemented "50% more bits per byte than the competition!" 12-bit fantasy console of my own invention - replete with inventions such as "UTF-12" - for shits and giggles.