performance - From explicit threading to TBB -
assuming have parallel algorithm uses explicit threading 1 or 2 locks synchronization , optimized take advantage of cache lines (including shared l3 cache between multiple cores), ways of incorporating tbb program? algorithm in question not break down nicely tasks threads.
without further information (code sample, or kind of generic presentation of algorithm, such flowchart) best way make such transition refactoring algorithm, isolating repeatable actions , trying combine them in tasks (a range of 1 or more activities have common purpose)...
unfortunately there's no magical formula make transition because these techniques fundamentally 2 answers solve same problem, therefore share common ground, tackle problem 2 different angles...
Comments
Post a Comment