←back to thread

218 points signa11 | 1 comments | | HN request time: 0.212s | source
Show context
pjmlp ◴[] No.43681194[source]
> The reason I believe C is and always will be important is that it stands in a class of its own as a mostly portable assembler language, offering similar levels of freedom.

When your computer is a PDP-11, otherwise it is a high level systems language like any other.

replies(7): >>43682173 #>>43682246 #>>43682554 #>>43682928 #>>43683332 #>>43683914 #>>43701871 #
grandempire ◴[] No.43682246[source]
Which other popular language more accurately represents a random access machine of fixed word length?
replies(3): >>43682309 #>>43682371 #>>43682763 #
kryptiskt ◴[] No.43682763[source]
C lacks sympathy with nearly all additions to hardware capabilities since the late 80s. And it's only with the addition of atomics that it earns the qualification of "nearly". The only thing that makes it appear as lower level than other languages is the lack of high-level abstraction capabilities, not any special affinity for the hardware.

For one, would expect that a low level language wouldn't be so completely worthless at bit twiddling. Another thing, if C is so low level, why can't I define a new calling convention optimized for my use case? Why doesn't C have a rich library for working with SIMD types that has been ubiquitous in processors for 25 years?

replies(4): >>43682820 #>>43682884 #>>43683999 #>>43684237 #
1. cv5005 ◴[] No.43684237[source]
simd doesnt make much sense as a standard feature/library for a general purpose language. If you're doing simd its because you're doing something particular for a particular machine and you want to leverage platform specific instructions, so thats why intrinsics (or hell, even externally linked blobs written in asm) is the way to go and C supports that just fine.

But sure, if all youre doing is dot products I guess you can write a standard function that will work on most simd platforms, but who cares, use a linalg library instead.