Designing high performance communication runtime for GPU managed memory: early experiences.
Dip Sankar BanerjeeKhaled HamidoucheDhabaleswar K. PandaPublished in: GPGPU@PPoPP (2016)
Keyphrases
- graphics processing units
- communication systems
- real time
- communication protocol
- state information
- case study
- general purpose
- multithreading
- communication networks
- graphics hardware
- memory usage
- data transfer
- interprocess communication
- gpu accelerated
- graphics processors
- past experience
- communication technologies
- parallel processing
- random access
- memory space
- low latency
- high efficiency
- communication cost
- tcp ip
- associative memory
- memory size
- gpu implementation
- memory requirements
- data structure
- compute intensive
- neural network