Compiler Support for Sparse Tensor Computations in MLIR
ACM Transactions on Architecture and Code Optimization2022Vol. 19(4), pp. 1–25
Citations Over TimeTop 10% of 2022 papers
Aart J. C. Bik, Penporn Koanantakool, Tatiana Shpeisman, Nicolas Vasilache, Bixia Zheng, Fredrik Kjølstad
Abstract
Sparse tensors arise in problems in science, engineering, machine learning, and data analytics. Programs that operate on such tensors can exploit sparsity to reduce storage requirements and computational time. Developing and maintaining sparse software by hand, however, is a complex and error-prone task. Therefore, we propose treating sparsity as a property of tensors, not a tedious implementation task, and letting a sparse compiler generate sparse code automatically from a sparsity-agnostic definition of the computation. This article discusses integrating this idea into MLIR.
Related Papers
- → Advanced loop optimizations for parallel computers(1988)24 cited
- → Comparative Cross-Platform Performance Results from a Parallelizing SML Compiler(2002)6 cited
- → An Interprocedural Framework for the Data and Loops Partitioning in the SIMD Machines(1999)
- → Evaluation of two compiler-based approaches for the parallelisation of an MPEG-2 decoder(1998)
- Comparative Cross-Platform Performance Results from a Parallelizing SML Compiler(2001)