Recursive DiamondCandy: non-memory-bound LRnLA algorithm for 3D cross stencil calculations on CUDA GPU.
Anastasia Y. PerepelkinaVadim D. LevchenkoPublished in: HP3C (2020)
Keyphrases
- parallel implementation
- experimental evaluation
- memory usage
- learning algorithm
- times faster
- gpu implementation
- memory requirements
- objective function
- computational cost
- worst case
- recognition algorithm
- preprocessing
- error bounds
- optimization algorithm
- high accuracy
- graphics processors
- dynamic programming
- computational complexity
- real time
- convergence rate
- memory space
- gpu accelerated
- general purpose
- np hard
- detection algorithm
- segmentation algorithm
- theoretical analysis
- expectation maximization
- particle swarm optimization
- recursive algorithm
- lower bound
- similarity measure
- decision trees