Programming for Hybrid Multi/Manycore MPP Systems by John Levesque Paperback Boo
The node itself will consist of 10's to 100's of MIMD (multiple instruction, multiple data) processing units with SIMD (single instruction, multiple data) parallel instructions. These new node architectures will represent a significant challenge to application developers.Programming for Hybrid Multi/Manycore MPP Systems attempts to briefly describe the current state-of-the-art in programming these systems, and proposes an approach for developing a performance-portable application that can effectively utilize all of these systems from a single application.
Jetzt bei Ebay: