This is the mail archive of the
mailing list for the GCC project.
Re: gcse, store motion and loop optimizer
- To: "Jan Hubicka" <jh at suse dot cz>, <gcc at gcc dot gnu dot org>, <dan at cgsoftware dot com>, <timothy dot c dot prince at intel dot com>
- Subject: Re: gcse, store motion and loop optimizer
- From: "Tim Prince" <tprince at computer dot org>
- Date: Fri, 3 Aug 2001 01:15:49 -0700
- References: <20010803095827.C10775@atrey.karlin.mff.cuni.cz>
----- Original Message -----
From: "Jan Hubicka" <firstname.lastname@example.org>
To: <email@example.com>; <firstname.lastname@example.org>;
Sent: Friday, August 03, 2001 12:58 AM
Subject: gcse, store motion and loop optimizer
> I am just tracking down the failure in twolf spec2000 benchmark. What
> happends there is basic-style programming, where all variables are
> and of course loop induction variable is too.
> Originally we did handled this in loop optimizer, by hoisting (We
> probably kill that code, but at least in twolf it still hoists, I
> re-check one Dan's store motion work is in) and optimized the loop
> in de-facto same was as if it were static.
> Now we do have problem. The induction variable is killed by gcse in
> steps, first it is PREed, then it is store motioned to sotre the final
> value, resulting in code, that use one regiester for iteration and
> always store the value to other register for storing the result.
> CSE is not able to unify these, as her local point of view suggest
> it is better to use first value for iterating (it don't know it is the
> induction variable) and then the second variable for storing (as the
> is longer and CSE decides it is superrior).
> This is unfortunate, as the strength reduction than goes crazy,
> redundant induction variable with value i+1 and doing other crazy
> confusing itself in the next loop iteration (by inserting instructions
> in the BIV computation where it don't expect) failing to unroll and
> ugly code. This is basically because it sees computation of
> separately and don't know, that it is just computation of the
> induction variable and that at the end there is induction=x
> (after several uses x and making several copies of it).
> Strength reduction has logic to get around some damage, but IMO it
> be able to expect that:
> 1) there are no dead computations of general induction variables
> 2) there are no unneeded copies
> as assuring those would require to do similar analyzis as gcse does.
> So my question is, would be possible to fix gcse pass to cleanup the
> damage? It can be done by eighter making store motion curefull and
> the value in register if it reaches, or run the copy propagation after
> it together with killing dead copy instructions (as subsequent CSE
> undoes copy prpagation partially currently). This would require CPROP
> to do the transformations even locally on the beggining and ends of
> blocks. I've experimented with using something like "REG_DEPRECATED"
> note to control CSE decisions, but I think it would be cleaner to
> gcse from being stricly global, as the GCSE itself isn't, just CPROP
> OK now I will continue to figure out what exactly goes wrong in the
> loop pass.
> Tim: May this be the case in your programs?
Making induction variables global is bad news. I'll check whether that
is happening in the samples I sent; if changing the source to make the
induction variables local solves the problem, I'll be satisified. I'd
even consider it good for the compiler to warn in -Wall -O mode if that
happens, and to indicate that it disables optimization. If it happens
in SPEC, I'm sure many compilers fail to optimize that. Does this
answer part of your question?