This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Scheduling for one processor variant, but reporting times for others
- From: Bradley Lucier <lucier at math dot purdue dot edu>
- To: Jim Wilson <wilson at specifixinc dot com>
- Cc: Bradley Lucier <lucier at math dot purdue dot edu>, gcc at gcc dot gnu dot org
- Date: Wed, 17 Sep 2003 10:45:04 -0500
- Subject: Re: Scheduling for one processor variant, but reporting times for others
On Wednesday, September 17, 2003, at 02:31 AM, Jim Wilson wrote:
Bradley Lucier wrote:
Would anyone be interested in implementing an option that reports,
via -dR, the cycle timing results for a CPU that the instructions
weren't scheduled for?
I think it would be easier to do this with a simulator or a profiling
tool.
Really? I know how to use gcc with -dR, I don't know how to use a
simulator.
This falls outside the scope of what the compiler is expected to do,
Perhaps.
and the gcc scheduler isn't designed to produce this info.
Well, you have two functions, one to order instructions "optimally" for
a given processor, and one to print the timing information for a given
processor given an instruction ordering. I don't see why the processor
in the first function has to be the same as the processor in the second
function.
I'm not going to continue to write e-mails asking for this, but you
raised what I consider to be a technical issue.
Brad