This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: cmov optimizations
- To: law at cygnus dot com, Richard Henderson <rth at cygnus dot com>
- Subject: Re: cmov optimizations
- From: Richard Henderson <rth at cygnus dot com>
- Date: Tue, 2 Jun 1998 22:05:25 -0700
- Cc: egcs-patches at cygnus dot com
- References: <19980524141219.17978@twiddle.rth.home> <7766.896848982@hurl.cygnus.com>
- Reply-To: Richard Henderson <rth at cygnus dot com>
On Tue, Jun 02, 1998 at 10:43:02PM -0600, Jeffrey A Law wrote:
> One of the other things what would be interesting would be to allow a
> more complex operand1 or operand2. They could be memory loads, shifts,
> or any other single instruction on the PA.
Are you talking about conditional move or conditional execution?
The difference in my mind being
load [mem],tmp
cmov pred,tmp,dest
vs
nullifynext pred
load [mem],dest
which does rather different things when [mem] is invalid. Similarly
for operations which might fault like div or any fp operation.
In light of this, it seems to me that you'd want to treat conditional
execution different from conditional move. Which is what I thought we
did. Which is confusing me about what you are describing here wrt
the PA.
> Looks reasonable to me.
In.
r~