On 2020-08-11, at 09:11:28, Jonathan Bradbury wrote:
> 
> I forgot to add that these instruction never actually get executed, no 
> exclusive-OR or subtraction is done.  Instead the instruction decoder 
> recognizes these cases and during register renaming which is part of 
> out-of-order execution, they map the target register to one that has a 
> constant value of 0 in it. It doesn't even take up an execution slot in the 
> pipeline.  Any subsequent instructions that depend on the 0 value will be 
> able to execute without having to wait for the computation to be done.
>  
Do they properly set CC?

Does setting CC possibly delay out-of-order execution of other
instructions which may set or test CC?

-- gil

Reply via email to