On Thu, Aug 25, 2016 at 09:11:56AM +0200, Vincent Guittot wrote: > On 10 August 2016 at 02:14, Yuyang Du <yuyang...@intel.com> wrote: > > In __decay_sum(), the 64-bit to 32-bit variable conversion makes no > > performance nor correctness use. > > Are you sure that there is no impact on 32bits system ? > Even this will be mainly shift manipulation > Honestly, I didn't consider 32-bit system at all, :(
And after all, it's 64-bit shift 32-bit vs. 64-bit shift 64-bit operations. I'm not an expert, but I intend to believe they are about the same, because the arithmetic overhead seems to lie in the left part of the shift operation, both of which are 64-bit (correct me if I'm wrong).
