RFR: 8013395 StringBuffer.toString performance regression impacting embedded benchmarks (original) (raw)

Remi Forax forax at univ-mlv.fr
Tue May 14 10:12:54 UTC 2013

On 05/14/2013 07:10 AM, David Holmes wrote:

[...]

So here is hopefully final webrev: http://cr.openjdk.java.net/~dholmes/8013395/webrev.v5/ It is the same approach as v3, but as Florian pointed out the cache should be cleared before the mutating action - just in case there is an exception. That leaves one issue that was flagged by a couple of folks: hotspot intrinsification of specific "string" usage patterns. I tracked this down in the hotspot code and I think it only applies in situations where the StringBuffer/StringBuilder could be elided completely - and so would not be an issue here. But I'm confirming this with the hotspot compiler folk (unfortunately the optimization is not clearly documented anywhere.) Thanks, David

thumb up for me.

Rémi

More information about the core-libs-dev mailing list