This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH][RFA] Split passes from gcse_main into separate passes
- From: Steven Bosscher <stevenb dot gcc at gmail dot com>
- To: Bradley Lucier <lucier at math dot purdue dot edu>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 27 Apr 2009 07:47:26 +0200
- Subject: Re: [PATCH][RFA] Split passes from gcse_main into separate passes
- References: <0D4A34C8-4FD9-4F86-A102-EC2E8828E28B@math.purdue.edu>
On Mon, Apr 27, 2009 at 12:21 AM, Bradley Lucier <firstname.lastname@example.org> wrote:
> With regard to:
>> After this split, we no longer have a reason to *not* alter jumps or
>> bypass jumps in CPROP. Therefore, the separate jump bypassing pass is
>> removed and CPROP now always alters jumps and bypasses jumps (this is
>> extremely cheap anyway).
> Would it be OK to remove this code in a separate patch, either before or
> after the addition of this new code?
That would be possible, but I'd have to add another rtl_opt_pass. So
I prefer not to do this.
> I've been trying to figure out how to analyze r118475, ?which removed "old"
> optimizations at the same time it added "new" optimizations, and it's hard
> to separate the two things figure out what has caused any slow-downs that
> one might see.
This isn't really a new optimization. In the old code, the jump_bypass
pass is just CPROP3 with alter jumps and bypass jumps. CPROP2 also
ran with alter jumps and bypass jumps. The only "new" bit is that we
now would do it in CPROP1 also. If something was slow here before,
you would have seen it already.