FALCON-X: Zero-copy MPI derived datatype processing on modern CPU and GPU architectures.
Jahanzeb Maqbool HashmiChing-Hsiang ChuSourav ChakrabortyMohammadreza BayatpourHari SubramoniDhabaleswar K. PandaPublished in: J. Parallel Distributed Comput. (2020)
Keyphrases
- heterogeneous computing
- parallel implementation
- graphics processing units
- real time
- parallel computing
- gpu implementation
- parallel architectures
- graphics processors
- general purpose
- high performance computing
- parallel programming
- parallel computation
- message passing interface
- parallel architecture
- multithreading
- processing units
- data transfer
- message passing
- parallelization strategy
- graphics hardware
- shared memory
- parallel processing
- memory hierarchy
- single instruction multiple data
- data processing