arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.15199cs.AR
收藏

Maple:一种基于行积的稀疏张量加速器处理单元

Maple: A Processing Element for Row-Wise Product Based Sparse Tensor Accelerators

Midia Reshadi, David Gregg

摘要:

稀疏张量计算是数据科学、图形处理和科学计算等领域中众多应用的核心计算部分。稀疏张量提供了跳过由零值引起的不必要计算的可能性。在本文中,我们提出了一种新的策略扩展行积稀疏张量加速器。我们提出了一种新的处理元素称为Maple,它使用多个乘法累积(MAC)单元来利用非零值的本地集群来增加并行性并减少数据移动。Maple采用压缩稀疏行(CSR)格式,并根据稀疏模式仅计算输入矩阵的非零元素。此外,我们可以使用Maple作为基于行积方法操作的各种空间张量加速器中的基本构建块。作为概念证明,我们在两个参考加速器中使用Maple:Extensor和Matraptor。我们的实验表明,在Matraptor和Extensor中使用Maple,与基线设计相比,分别实现了50%和60%的能量效益和15%和22%的加速比。与基线结构相比,采用Maple还分别使Matraptor和Extensor的面积消耗减少了5.9倍和15.5倍。

英文摘要:

Sparse tensor computing is a core computational part of numerous applications in areas such as data science, graph processing, and scientific computing. Sparse tensors offer the potential of skipping unnecessary computations caused by zero values. In this paper, we propose a new strategy for extending row-wise product sparse tensor accelerators. We propose a new processing element called Maple that uses multiple multiply-accumulate (MAC) units to exploit local clusters of non-zero values to increase parallelism and reduce data movement. Maple works on the compressed sparse row (CSR) format and calculates only non-zero elements of the input matrices based on the sparsity pattern. Furthermore, we may employ Maple as a basic building block in a variety of spatial tensor accelerators that operate based on a row-wise product approach. As a proof of concept, we utilize Maple in two reference accelerators: Extensor and Matraptor. Our experiments show that using Maple in Matraptor and Extensor achieves 50% and 60% energy benefit and 15% and 22% speedup over the baseline designs, respectively. Employing Maple also results in 5.9x and 15.5x smaller area consumption in Matraptor and Extensor compared with the baseline structures, respectively.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑