Distributed out-of-memory NMF on CPU/GPU architectures.
Ismael BoureimaManish BhattaraiMaksim Ekin ErenErik SkauPhilip RomeroStephan J. EidenbenzBoian S. AlexandrovPublished in: J. Supercomput. (2024)
Keyphrases
- data transfer
- heterogeneous computing
- memory hierarchy
- graphics processors
- graphics processing units
- compute intensive
- single instruction multiple data
- memory bandwidth
- multithreading
- negative matrix factorization
- intel xeon
- cooperative
- memory management
- parallel computers
- heterogeneous environments
- real time
- distributed systems
- gpu implementation
- data access
- commodity hardware
- matrix factorisation
- nonnegative matrix factorization
- memory access
- distributed environment
- computing power
- parallel architectures
- data management
- dynamically created
- memory requirements
- main memory
- level parallelism
- parallel computation
- multi core processors
- computing environments
- parallel computing
- limited memory