mGEMM: low-latency convolution with minimal memory overhead optimized for mobile devices

Jongseok Park, Kyungmin Bin, Kyunghan Lee. mGEMM: low-latency convolution with minimal memory overhead optimized for mobile devices. In Nirupama Bulusu, Ehsan Aryafar, Aruna Balasubramanian, Junehwa Song, editors, MobiSys '22: The 20th Annual International Conference on Mobile Systems, Applications and Services, Portland, Oregon, 27 June 2022 - 1 July 2022. pages 222-234, ACM, 2022. [doi]

Abstract

Abstract is missing.