Michael Santy wrote: > >> At Fri, 05 Mar 2010 09:10:19 +0100, Martin Sustrik wrote: >> >> This is possible, if language runtime supports callbacks from foreign >> thread (to call language code from C). However, handwritten memory >> management wrapper kills performance. Unless you have to deal with >> very large messages, it's usually faster just to copy data. > > Thanks -- that definitely makes sense. I agree that this would not be > appropriate for small messages. The messages that I'm considering creating > are on the order of 200MB and are scattered across a handful of contiguous > regions of memory.
In that case zero-copy makes sense IMO. As far as I understand the point where copying becomes problematic is when CPU cache size - which is couple of MB in most cases - is exceeded. Martin _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
