By: Gabriele Svelto (gabriele.svelto.delete@this.gmail.com), February 14, 2013 1:37 am
Room: Moderated Discussions
Maynard Handley (name99.delete@this.redheron.com) on February 13, 2013 3:53 pm wrote:
> The right solution is what it has always been --- the PPC/POWER solution.
> (a) Multiple condition registers.
> (b) Most instructions come in two variants, one of which sets conditions, one of which does not.
> (c) Compares can choose which condition register to modify.
The PPC/POWER implementation of this feature has a significant technical drawback for an implementer though: the fact that you can access the condition register with different granularities. Either as a whole 32-bit register, as 8 4-bit fields or on a bit-by-bit basis. This makes renaming it a much more complicated task than it needs to be (and is the reason why some of the instructions accessing it end up being slower than others in many implementations).
> The right solution is what it has always been --- the PPC/POWER solution.
> (a) Multiple condition registers.
> (b) Most instructions come in two variants, one of which sets conditions, one of which does not.
> (c) Compares can choose which condition register to modify.
The PPC/POWER implementation of this feature has a significant technical drawback for an implementer though: the fact that you can access the condition register with different granularities. Either as a whole 32-bit register, as 8 4-bit fields or on a bit-by-bit basis. This makes renaming it a much more complicated task than it needs to be (and is the reason why some of the instructions accessing it end up being slower than others in many implementations).
