[llvm] r219135 - [BasicAA] Revert "Revert r218714 - Make better use of zext and sign information."

Hal Finkel hfinkel at anl.gov
Thu Nov 13 01:20:48 PST 2014


After some more review and updates from Nick, trying again in r221876.

 -Hal

----- Original Message -----
> From: "Nicholas White" <n.j.white at gmail.com>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: "Commit Messages and Patches for LLVM" <llvm-commits at cs.uiuc.edu>, "Lang Hames" <lhames at gmail.com>, "James
> Molloy" <James.Molloy at arm.com>, "Richard Barton" <richard.barton at arm.com>
> Sent: Tuesday, October 21, 2014 4:06:10 AM
> Subject: Re: [llvm] r219135 - [BasicAA] Revert "Revert r218714 - Make better use of zext and sign information."
> 
> 
> 
> OK, I've uploaded a revised diff to http://reviews.llvm.org/D5866 .
> It now skips the expensive ComputeSignBit logic when AllPositive is
> false. It still has to evaluate isValueEqualInPotentialCycles for
> every GEPVariable regardless of AllPositive as the modulo analysis
> is only safe (i.e. correct) if all the variables don't change across
> cycles. Thanks -
> 
> Nick
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory



More information about the llvm-commits mailing list