POPA: Expressing High and Portable Performance across Spatial and Vector Architectures for Tensor Computations
Xiaochen Hao, Hongbo Rong, Mingzhe Zhang, Ce Sun, Hong Jiang, Yun Liang · 2024
This paper aims at high and portable performance for tensor computations across spatial (e.g., FPGAs) and vector architectures (e.g., GPUs). The state-of-the-art usually address performance portability across vector architectures (CPUs and GPUs). However, they either miss FPGAs or do not achieve high performance. Without a common architectural abstraction, they program and optimize spatial and vector devices separately, causing low portability.