RFR: AARCH64: optimize string compare intrinsic

Dmitrij Pochepko dmitrij.pochepko at bell-sw.com
Fri May 4 13:24:44 UTC 2018

On 04.05.2018 15:59, Andrew Haley wrote:
> On 05/04/2018 12:28 PM, Dmitrij Pochepko wrote:
>> vectorizedMismatch intrinsic contract moves type information from
>> compile time to runtime (to have single entry point),
> It does not have to do that: there's no reason not to be able to
> use the same vectorized-mismatch-generating logic able to use
> constant information (i.e. the type) conditionally at code gen
> time.
Do you suggest to change vectorizedMismatch from generic single entry 
point to 4 versions (1,2,4 and 8 -byte) each optimized for respective 
size(and possible re-using code generation logic)? Then it can be 
re-used for same-encoded strings without penalties, indeed, but it 
requires changes in jdk.internal.util.ArraysSupport.java and also 
updating existing vectorizedMismatch intrinsic implementation on x86.

This is definitely doable but looks like a separate cross-platform effort.

More information about the hotspot-compiler-dev mailing list