Andrei Alexandrescu Wrote: > http://www.reddit.com/r/programming/comments/cmxlj/concurrency_in_the_d_programming_language/ > > Andrei
Wonderful Andrei. Great information and a fine gift. Must buy book. BTW: While you were writing that chapter I was busy writing a multi-core implementation of my Dendrite flow-based programming system for D/Tango which is based on a high-performance, lock-free message passing design. In fact, I'm still writing it :-) Initial tests indicate core-to-core (inter-thread) message passing is running at around 20 million messages / second, lock-free with no thread de-scheduling. That's probably best case, YMMV etc. Lock-free FIFO is based on an optimized, batching, cache-coherent algorithm invented by someone for content sniffing pipelines. I guess it's like anything on the Internet. If it's worth doing, at least three people are thinking about it and 2 people are doing it.
