Automatic fusions of CUDA-GPU kernels for parallel map.
Jan FousekJiri FilipovicMatus MadzinPublished in: SIGARCH Comput. Archit. News (2011)
Keyphrases
- parallel implementation
- parallel computing
- parallel computation
- parallel programming
- compute unified device architecture
- graphics processing units
- gpu implementation
- shared memory
- graphics hardware
- graphics processors
- parallel processing
- gpu accelerated
- parallel algorithm
- general purpose
- parallel hardware
- distributed memory
- cpu implementation
- kernel methods
- support vector
- feature maps
- memory bandwidth
- single instruction multiple data
- cluster of workstations
- parallel computers
- computer architecture
- real time
- multiple kernel learning
- maximum a posteriori
- computer graphics
- feature space
- multiscale
- machine learning
- data sets