Abstract
The mapping of tensor computation is a complex and important process for spatial accelerators. Today's mapping works depend on hand-tuned kernel libraries or search-based heuristics from human experts. The former is time-intensive while the latter easily leads to sub-optimal performance. In this paper, we propose TensorMap, a deep reinforcement learning (RL)-based mapping framework for tensor computations on spatial accelerators. We propose a sequential generation mode for mapping optimization and construct a coarse-grained action space to reduce the complexity of the mapping search space. An efficient policy network is devised to optimize mapping primitives in the RL-based search. We then propose a stop signal that is sampled from Bernoulli distribution to facilitate multi-level loop unrolling for spatial accelerators. Finally, a genetic algorithm is employed to further refine the optimized mappings. In the experiments, we demonstrate TensorMap's ability for different spatial accelerators with various tensor computations. On TPU, TensorMap provides 2.6 $\times$ , 2.7 $\times$ , and 2.4 $\times$ better energy-delay product (EDP) on average compared with FlexTensor, Ansor, and AMOS respectively. On Eyeriss, TensorMap provides 2.1 $\times$ , 1.8 $\times$ , and 1.7 $\times$ better EDP on average compared with FlexTensor, Ansor, and AMOS respectively.
Funding Information
  • National Key R&D Program of China (2022ZD0115304)
  • Natural Science Foundation of China (62072479, 62332021)
  • Major Program of Guangdong Basic and Applied Research (2019B030302002)
  • Natural Science Foundation of Guangdong Province (2024B1515020011)
  • Guangzhou Science and Technology Projects (202201011388)
  • Xiaomi Young Talents Program
  • Pazhou Lab (PZL2023KF0001)