Re: Performance assistance / advice

On Mar 12, 2025, at 8:51 AM, Tim Bradshaw (as tfb at cley dot com) <[email protected]> wrote:
> 
> Every new feature of a processor will absolutely have an interface to it which can be used from C.

As someone who has made a career of programming the metal in C, this statement really hasn't been true up until this point, so I'm perplexed why you think it would be true in the future.  C simply gets around this problem by having a lightweight interface to produce inline non-portable native assembly instructions.

One simple case in point where this is needed in low-level systems code are barrier instructions in the ARMv8 instruction set.  The SIMD operations in every modern processor are another great example of where your claim about the level of machine-specific support for C doesn't really match up with reality.

Now, that said, if one reads your claims in relative terms compared to what support might be available in Lisp dialects compared to C dialects, I suppose one could argue for your position as a legitimate "vibe" even if your point is flawed in an absolute rigid sense.

-- Tahoma

P.S. Don't get my intent wrong here.  I'm much more of a fan of Lisp than I ever will be of C, even if I've made a profession out of mastery of the latter.  Regardless, your statement didn't really match up with my decades of experience of being the type of C developer who directly programs machine architectural features, so I felt obliged to comment.