Conjugate gradient method with graphics processing unit acceleration: CUDA vs OpenCL.
Abal-Kassim Cheik AhamedFrédéric MagoulèsPublished in: Adv. Eng. Softw. (2017)
Keyphrases
- conjugate gradient method
- graphics processing units
- compute unified device architecture
- gpu implementation
- general purpose
- shared memory
- parallel implementation
- parallel computing
- graphics processors
- graphics hardware
- parallel algorithm
- parallel programming
- conjugate gradient
- parallel computation
- parallel processing
- cpu implementation
- objective function
- positive definite
- computing systems
- convergence property
- massively parallel
- efficient implementation
- iterative methods
- real time
- floating point
- message passing
- processing units
- high performance computing
- parallel architectures
- neural network
- covariance matrix
- maximum likelihood estimation
- training algorithm
- least squares
- general purpose computing