Optimizing Non-commutative Allreduce Over Virtualized, Migratable MPI Ranks.
Sam WhiteLaxmikant V. KaléPublished in: IPDPS Workshops (2022)
Keyphrases
- parallel algorithm
- cloud computing
- message passing
- general purpose
- parallel implementation
- shared memory
- data center
- parallel computing
- parallelization strategy
- real time
- rank order
- high performance computing
- data sets
- resource management
- ranked list
- ranking functions
- expert systems
- information retrieval
- databases
- message passing interface