4 ms·
In the linked slides, DJB talks about a language for communication with the compiler, separating optimizations from specification. This reminded me of VPRI's "M
by fovc 9y ago
In the linked slides, DJB talks about a language for communication with the compiler, separating optimizations from specification. This reminded me of VPRI's "Meaning separated from optimization" [1] principle. Does anyone know what became of that line of thinking? Is this idea making it's way into Ohm? I remember reading a post/paper about optimizing Nile/Gezira to better exploit the CPU cache (and the struggle to use SIMD), but can't seem to find it now.
[1] http://www.vpri.org/pdf/rn2006002_nsfprop.pdf http://www.vpri.org/pdf/rn2006002_nsfprop.pdf
- nickpsecurity 9y agoIBM's old PL/S language allowed you to give hints like where data would go or what checks would happen right in the function declaration. The compiler would handle it from there.