I'd imagine that he was talking about cache invalidation protocols in (massively) multiprocessor computers. This is indeed a very difficult thing to get right without destroying performance by sending too many bus requests. (I've just spent two evenings reading about the problem...)
I'd imagine that he was talking about cache invalidation protocols in (massively) multiprocessor computers. This is indeed a very difficult thing to get right without destroying performance by sending too many bus requests. (I've just spent two evenings reading about the problem...)
Rich - 5th July 2007 06:55 - #