Re: Long.bitCount micro-optimization

[Maurizio Cimadamore]

To expand a bit on Martin's point - I think your benchmark should 
include a 'baseline' version of bitCountInt/bitCountLong whose 
implementation simply delegates to the JDK methods 
(Integer.bitCount/Long.bitCount) - I see that bitCount is on the list of 
supported intrinsics [1], so, by adding those 'baselines' 
intrinsification should definitively kick in, and give you different 
(possibly even better) numbers.

[1] - 
http://hg.openjdk.java.net/jdk9/dev/hotspot/file/tip/src/share/vm/classfile/vmSymbols.hpp#l830

Maurizio


On 09/05/17 03:54, Martin Buchholz wrote:
I would also have benchmarked against the intrinsified version to see how
it did.