Skip to main content

You know, I wonder how much latency and performance overhead you could reduce by dynamically generating code for and recompiling parts of the audio server according to your system configuration. Generally, I wonder how much more efficient our computers could be if systems-level components were created in a language with the rich metaprogramming capabilities of Lisp. Would the opportunities for dynamic recompilation and optimization outweigh the costs associated with such a runtime? And how could we reduce or eliminate those costs?
in reply to Urusan

Urusan
Urusan  
•
As someone who has done a decent amount of work in a language where this idea is possible, the biggest difficulty will be that you may need to stop and compile at any time. In particular, if you want the most efficient code, the compiler may need more time than is possible with JIT compilation. That said, I do think this issue is solvable, and Julia is making good progress on it.

A more fundamental problem is the need for garbage collection. This is a problem for systems-level stuff.
⇧