Optimizing half precision Winograd convolution on ARM many-core processors.
Dedong XieZhen JiaZili ZhangXin JinPublished in: APSys (2022)
Keyphrases
- mesh connected
- high precision
- image processing
- parallel algorithm
- parallel processing
- multiscale
- precision and recall
- average precision
- high recall
- high performance computing
- convolution kernel
- database
- embedded processors
- high end
- parallel computation
- parallel computing
- binary images
- computer vision
- genetic algorithm