Interconnection of processors. Topologies of networks of shared environment, direct networks and indirect networks. Design of multicomputers (distributed memory). Methods of commutation, blocking control, routing algorithms. Design of multiprocessors (shared memory). Cache coherence, coherence algorithms in shared environment networks and coherence algorithms based on directory. Segmented processors. Data dependencies (RAW, WAW, WAR) structural, data and control conflicts and out-of-order completion. Supersegmented processors. Structural, data and control conflicts, condition prediction, branch target prediction (BTB). Superscalar processors. Parallelism, out-of-order completion, dynamic assignment of registers, interruptions and breaks in sequence.History buffer, reorder buffer, etc. Access to memory.Coherence between cache memory and main memory, synchronization. An architecture.Speculative architectures, TLP and ILP, speculative processors. Clusters, grids and cloud computing. Programming for messages, cluster environment MOSIX, performance.
|
OBCompulsory |
5.00 |
A |
1first semester |