Using hardware performance counters to speed up autotuning convergence on GPUs.
Jiri FilipovicJana HozzováAmin NezaratJaroslav OlhaFilip PetrovicPublished in: J. Parallel Distributed Comput. (2022)
Keyphrases
- graphics processing units
- graphics hardware
- graphics cards
- low cost
- parallel architectures
- digital signal processing
- real time
- computational power
- general purpose
- convergence rate
- graphics processors
- convergence speed
- hardware and software
- floating point
- parallel hardware
- computing power
- commodity hardware
- image processing
- high end
- heterogeneous computing
- hardware architecture
- massively parallel
- data acquisition
- neural network
- parallel programming
- parallel computation
- highly parallel
- hardware implementation
- signal processing
- high speed