This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [patches] Re: SSE min/max instruction support


> On Sat, Feb 24, 2001 at 02:20:13AM +0100, Jan Hubicka wrote:
> > 	* i386.md (minsf*, mindf*, maxsf*, maxdf*): New instruction patterns and
> > 	splitters.
> > 	* i386.c (ix86_expand_fp_movcc): Recognize min and max instructions.
> 
> Any reason not to use smin/smax rtx codes?
`(smin:M X Y)'
`(smax:M X Y)'
     Represents the smaller (for `smin') or larger (for `smax') of X
     and Y, interpreted as signed integers in mode M.
Don't seems to be ready for floats.
There is no infrastructure in simplify-rtx/combine.c to handle these, so
I came to conclusion that it is better to not use it.
I see no advantage of this construct over the if_then_else (after possible
combine tweek to canonize equivalent if_then_elses to the one mathced
by my patterns).
But maybe I am mistaken - if you believe it is better, I will use smin/smax
and add the neccesary infrastructure bits too.

Honza
> 
> 
> r~


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]