Memory-Constrained Vectorization and Scheduling of Dataflow Graphs for Hybrid CPU-GPU Platforms.
Shuoxin LinJiahao WuShuvra S. BhattacharyyaPublished in: ACM Trans. Embed. Comput. Syst. (2018)
Keyphrases
- heterogeneous computing
- parallel computing
- graphics processors
- compute intensive
- multithreading
- intel xeon
- memory bandwidth
- graphics processing units
- data transfer
- graph transformation
- gpu implementation
- computing platform
- data flow
- limited memory
- level parallelism
- scheduling problem
- memory hierarchy
- real time
- series parallel
- multi threaded
- memory access
- resource consumption
- control flow
- parallel machines
- scheduling algorithm
- general purpose
- parallel processors
- parallel computation
- memory requirements
- computing systems
- parallel programming
- graphics hardware
- high performance computing
- computational power
- parallel implementation
- directed graph
- single instruction multiple data
- line drawings
- parallel processing
- data structure
- np hard
- main memory
- weighted graph
- graph theory
- graph matching