[PATCH] D20172: [AArch64] Disable narrow load merge by default

[James Molloy via llvm-commits]

Hi Jun,

Have you considered deciding this as a MachineCombiner pattern? This would be a good place to know if the loop is arithmetic or load/store heavy.

Cheers,

James
> On 11 May 2016, at 16:51, Jun Bum Lim <junbuml at codeaurora.org> wrote:
> 
> junbuml added a comment.
> 
> No performance regression was found in spec200/2006.  I will run EEMBC as well.
> 
> 
> http://reviews.llvm.org/D20172
> 
> 
>