[concurrency-interest] A Fast Wait-Free HashTable

Doug Lea dl at cs.oswego.edu
Wed Feb 21 09:20:16 EST 2007


Peter Veentjer wrote:
> Does anyone know if this structure supports 'safe handoff'?
> 

The j.u.c versions most certainly will. And I believe the
one the slides report performance results on also does.

(The "extra support in atomics package" I mentioned
is mainly to make generally available operations
that allow you to more carefully control these kinds of
happens-before relations that external users need without
adding needless overhead to internal aspects that don't.
Cliff needed to use underlying VM primitives to get
this effect. We'll be adding some classes/operations
that expose these to non-privileged programs, so anyone
can use them.)

-Doug


> On 2/21/07, Doug Lea <dl at cs.oswego.edu> wrote:
>> David J. Biesack wrote:
>> > Does anyone know more about Cliff Click's wait-free hashtable?
>> >
>> > http://cs.stanford.edu/calendar/abstract.php?eventId=2270
>> >


More information about the Concurrency-interest mailing list