Cost-driven thread coarsening for GPU kernels.
Prithayan BaruaJun ShirakoVivek SarkarPublished in: PACT (2018)
Keyphrases
- data driven
- real time
- high cost
- data sets
- cost reduction
- minimum cost
- total cost
- multiple kernel learning
- cost savings
- genetic algorithm
- gaussian kernels
- graphics processors
- kernel function
- kernel methods
- cost sensitive
- supply chain
- support vector
- parallel implementation
- data structure
- expected cost
- graphics hardware
- finer granularity