GAMMA: Automating the HW Mapping of DNN Models on Accelerators via Genetic Algorithm
GAMMA: Automating the HW Mapping of DNN Models on Accelerators via Genetic Algorithm
复制标题
DOI：
10.1145/3400302.3415639
复制
发表时间：
2020-11
期刊：
影响因子：
--
通讯作者：
Sheng-Chun Kao;T. Krishna
中科院分区：
文献类型：
--
作者：
Sheng-Chun Kao;T. Krishna
DNN 层是多维循环，可以在 DNN 加速器上跨空间和时间以多种方式进行排序、平铺和调度。这些选择中的每一个都称为映射。事实证明，映射在整体性能和效率中起着极其重要的作用，因为它直接决定了加速器可以从 DNN 中利用的重用量。此外，研究揭示了优化每层映射的好处，而不是为每个 DNN 层使用固定映射。然而，在给定加速器和层的情况下确定正确的映射仍然是一个悬而未决的问题。巨大的映射空间（或映射空间）使得暴力穷举搜索方法难以实现。在本文中，我们提出了一种基于领域特定遗传算法的方法 GAMMA，该方法是专门为这种硬件映射问题而设计的。与之前的工作相比，要么针对具有有限映射空间的简单刚性加速器，要么从有限的映射集中进行选择，我们构建了一个极其灵活的映射空间，并表明 GAMMA 可以探索该空间并确定具有高采样效率的优化映射。我们将 GAMMA 与许多流行的优化方法进行定量比较，并观察 GAMMA 不断找到更好的解决方案。
DNN layers are multi-dimensional loops that can be ordered, tiled, and scheduled in myriad ways across space and time on DNN accelerators. Each of these choices is called a mapping. It has been shown that the mapping plays an extremely crucial role in overall performance and efficiency, as it directly determines the amount of reuse that the accelerator can leverage from the DNN. Moreover, instead of using a fixed mapping for every DNN layer, research has revealed the benefit of optimizing per-layer mappings. However, determining the right mapping, given an accelerator and layer is still an open question. The immense space of mappings (or map-space) makes brute-forced exhaustive search methods unapproachable. In this paper, we propose a domain-specific genetic algorithm-based method, GAMMA, which is specially designed for this HW-mapping problem. In contrast to prior works that either target simple rigid accelerators with a limited map-space or choose from a restricted set of mappings, we construct an extremely flexible map-space and show that GAMMA can explore the space and determine an optimized mapping with high sample efficiency. We quantitatively compare GAMMA with many popular optimization methods and observe GAMMA consistently finds better solutions.