3 ms·
I'd like to see this same technique applied to x86, and what the performance is like without the "illegal instructions" (omitting them from generation would pro
by userbinator 4y ago
I'd like to see this same technique applied to x86, and what the performance is like without the "illegal instructions" (omitting them from generation would probably be trivial). It's relatively well known that one of the ways Asm programmers can beat compilers is on instruction selection, and that's what this technique seems to excel at.