An Intermediate Language for General Sparse Format Customization
An Intermediate Language for General Sparse Format Customization
复制标题
DOI：
10.1109/lca.2023.3262610
复制
发表时间：
2023-07
影响因子：
2.3
通讯作者：
Jie Liu-;Zhongyuan Zhao;Zijian Ding;Benjamin Brock;Hongbo Rong;Zhiru Zhang
中科院分区：
文献类型：
--
作者：
Jie Liu-;Zhongyuan Zhao;Zijian Ding;Benjamin Brock;Hongbo Rong;Zhiru Zhang
硬件专业化的必然趋势促使在处理稀疏工作负载（通常受内存限制）时越来越多地使用自定义数据格式。这些格式有助于自动生成具有目标感知的数据布局，以改善内存访问延迟和带宽利用率。然而，现有的稀疏张量编程模型和编译器对高效定制稀疏格式几乎没有提供支持。此外，由于这些框架采用基于属性的方法进行格式抽象，它们无法轻易扩展以支持通用格式定制。为了克服这一缺陷，我们提出了UniSparse，这是一种中间语言，它为表示和定制稀疏格式提供了统一的抽象。我们还开发了一个利用MLIR基础设施的编译器，它支持格式的自适应定制。我们通过在多个不同的硬件目标（包括英特尔CPU、英伟达GPU和模拟的内存处理（PIM）设备）上使用混合格式运行常用的稀疏线性代数运算的实验，证明了我们方法的有效性。
The inevitable trend of hardware specialization drives an increasing use of custom data formats in processing sparse workloads, which are typically memory-bound. These formats facilitate the automated generation of target-aware data layouts to improve memory access latency and bandwidth utilization. However, existing sparse tensor programming models and compilers offer little or no support for productively customizing the sparse formats. Moreover, since these frameworks adopt an attribute-based approach for format abstraction, they cannot easily be extended to support general format customization. To overcome this deficiency, we propose UniSparse, an intermediate language that provides a unified abstraction for representing and customizing sparse formats. We also develop a compiler leveraging the MLIR infrastructure, which supports adaptive customization of formats. We demonstrate the efficacy of our approach through experiments running commonly-used sparse linear algebra operations with hybrid formats on multiple different hardware targets, including an Intel CPU, an NVIDIA GPU, and a simulated processing-in-memory (PIM) device.