Parallel DSP algorithms on TurboNet: an experimental system with hybrid message-passing/shared-memory architecture.
Xi LiSotirios G. ZiavrasConstantine N. ManikopoulosPublished in: Concurr. Pract. Exp. (1996)
Keyphrases
- shared memory
- message passing
- multi processor
- graphic processing unit
- parallel architectures
- parallel computers
- parallel architecture
- distributed memory
- sum product algorithm
- belief propagation
- parallel computing
- parallel programming
- parallel algorithm
- inference in graphical models
- matrix multiplication
- parallel machines
- max product
- distributed systems
- parallel computation
- multithreading
- distributed shared memory
- single processor
- loopy belief propagation
- parallel execution
- factor graphs
- message passing interface
- memory access
- efficient implementation
- bayesian networks
- parallel tree search
- shared memory multiprocessors
- sum product
- multi core systems
- shared memory multiprocessor
- commodity hardware
- parallel implementation
- graphical models
- markov random field
- lower bound