Classifying and alleviating the communication overheads in matrix computations on large-scale NUMA multiprocessors.
Yi-Min WangHsiao-Hsi WangRuei-Chuan ChangPublished in: J. Syst. Softw. (1998)
Keyphrases
- load balancing
- shared memory multiprocessors
- matrix multiplication
- response time
- distributed memory
- linear algebra
- communication networks
- real world
- communication overhead
- automatic classification
- communication systems
- shared memory
- communication cost
- small scale
- low rank
- wireless networks
- low rank matrix approximation
- real life
- information exchange
- matrix factorization
- parallel implementation
- multi party
- singular value decomposition
- data acquisition
- information sharing
- communication protocol
- execution model