RFR (XS): CR 8006176: Switch to optimal identity hash code generator

Doug Lea dl at cs.oswego.edu
Mon Jan 14 06:41:48 PST 2013

Aleksey: I'm not subscribed to this list, so please forward if you
don't see it gated:

On 01/14/13 09:21, Aleksey Shipilev wrote:
> In related news, ...
> http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2013-January/005212.html

My take is that the only possible concern is statistical
independence of the thread-local versions? But this is pretty
easy to address if it hasn't already been done.
(I had to do this in CR 8005926 for ThreadLocalRandom.)
Use an independent generator for their initial seeds. And
if necessary (which it was for ThreadLocalRandom) further
force independence from a global sequence merely by starting
the TL version in a different place (I did this using a program
that advanced the common seed a few billion times, and
then used that value.)


More information about the hotspot-runtime-dev mailing list