[llvm-commits] [llvm-testresults] lab-mini-01__O3-plain__clang_DEV__x86_64 test results

[Hal Finkel]

On Fri, 25 May 2012 20:13:46 -0700
Andrew Trick <atrick at apple.com> wrote:

> 
> 
> On May 25, 2012, at 7:38 PM, Hal Finkel <hfinkel at anl.gov> wrote:
> 
> > What do you mean by 'clearly'? We'd need to define some metric for
> > this, and I'm not sure what that should be.
> 
> As long as you can unambiguously define which forms are more/less
> canonical and show that the more canonical form may expose a useful
> transformation, then it's reasonable. I'm not looking for a
> cost/benefit analysis. 

Okay.

> 
> In Duncan's case I could see no reason for reassociation.

If one of the purposes of the canonical forms is to assist with CSE
(and GVN, etc.), then it does have a potential purpose, just one not
realized in this example.

 -Hal

> 
> Andy



-- 
Hal Finkel
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory