Author: Robert Hyatt
Date: 12:58:47 07/22/03
Go up one level in this thread
BTW, the nullification stuff has been around forever. Sun did it to solve the problem of fetching instruction N+1 while still decoding instruction N to notice it is a branch. It allows them to fetch one instruction beyond the branch _safely_ because it is always executed. Unless you use the nullify conditional to say "execute it if the branch is taken, but don't execute it if we fall through." I've written a lot of SPARC assembly, and I always found a way to avoid using the nullification hogwash to avoid an empty slot in the instruction pipeline. Other architectures have used "skip" type instructions, ie compare and skip next if comparison is true. Even in 1970 Bell South taught us to do self-modifying code to do this. IE add a constant to an instruction op-code to turn it into a noop where it was a move, etc. Ugly, of course. :)
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.