Characterizing CUDA Unified Memory (UM)-Aware MPI Designs on Modern GPU Architectures.
Karthik Vadambacheri ManianA. A. AmmarAmit RuhelaChing-Hsiang ChuHari SubramoniDhabaleswar K. PandaPublished in: GPGPU@ASPLOS (2019)
Keyphrases
- parallel implementation
- parallel computing
- parallel programming
- parallel computers
- graphics processors
- shared memory
- parallel computation
- parallel algorithm
- compute unified device architecture
- message passing interface
- multi core processors
- gpu implementation
- heterogeneous computing
- memory bandwidth
- memory management
- graphics hardware
- compute intensive
- parallel architectures
- gpu accelerated
- memory requirements
- processing elements
- general purpose
- parallel processing
- multithreading
- graphic processing unit
- memory hierarchy
- memory access
- distributed memory
- high performance computing
- graphics processing units
- massively parallel
- multi threaded
- computing power
- computational power
- parallelization strategy
- real time
- single instruction multiple data
- digital signal processors
- message passing
- memory size
- memory usage
- computing systems
- hardware implementation
- design tools
- field programmable gate array
- cloud computing
- parallel execution
- computer architecture
- programming environment
- memory space