[concurrency-interest] Queue quest
nitsanw at yahoo.com
Tue Apr 8 13:13:52 EDT 2014
I'm busy building a lock free queue library: https://github.com/nitsanw/JAQ-InABox
It's early stages, an I admit to not having what you are after at this point. The MPSC bounded queue is allot faster than the JDK variety but it is an array backed one to the bounded size. I think I can see a way to match your requirement but perhaps we can take it off list so I can have more detail.
On Tuesday, April 8, 2014 6:25 PM, √iktor Ҡlang <viktor.klang at gmail.com> wrote:
On Tue, Apr 8, 2014 at 6:07 PM, Sergey Mashkov <comm at sergey-mashkov.net> wrote:
>Well, if it will not make you anger, why don't you mention Java's LinkedBlockingQueue? In 99% cases it's sufficient so you generally don't need lock-free algos.
Yep, that's a good fallback, but I figured I'd ask first :)
>By the way I don't think it's possible to implement in practice actually efficient non-blocking queue with no preallocated buffer.
Well, for some definition of efficient :-)
Given that most queues spend their lives either full or empty, a preallocated buffer may not make sense in case there are many millions of them and they spend their lives mostly empty.
>√iktor Ҡlang писал 2014-04-08 06:38:
>>I thought I'd throw this question out there before I go all out NIH.
>>Does anybody know of an open source (apache 2 compatible) "minimal
>>overhead", non-blocking, bounded, non-constant-space (i.e. no
>>ringbuffer or preallocated size array) multiple-producer
>>single-consumer/multiple-consumer queues in Java/bytecode?
>>Concurrency-interest mailing list
>>Concurrency-interest at cs.oswego.edu
>Concurrency-interest mailing list
>Concurrency-interest at cs.oswego.edu
Concurrency-interest mailing list
Concurrency-interest at cs.oswego.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Concurrency-interest