Vectorizing Sparse Matrix Codes with Dependency Driven Trace Analysis

TL;DR

This paper introduces PSC, a polyhedral model enabling vectorization of irregular sparse matrix computations, significantly improving performance over existing methods through a novel dependency-driven trace analysis framework.

Contribution

The paper presents PSC, a new polyhedral model for vectorizing irregular sparse computations, and a framework that efficiently mines memory accesses to generate high-performance code.

Findings

DDF outperforms MKL with 1.93x and 4.5x speedups for SpMV and SpTRSV.

DDF outperforms Sympiler by up to 11x for SpTRSV.

DDF outperforms Augustine et al. by up to 12x for SpMV.

Abstract

Sparse computations frequently appear in scientific simulations and the performance of these simulations rely heavily on the optimization of the sparse codes. The compact data structures and irregular computation patterns in sparse matrix computations introduce challenges to vectorizing these codes. Available approaches primarily vectorize regular regions of computations in the sparse code. They also reorganize data and computations, at a cost, to increase the number of regular regions. In this work, we propose a novel polyhedral model, called the partially strided codelets (PSC), that enables the vectorization of computation regions with irregular data access patterns. PSCs also improve data locality in sparse computation. Our DDF inspector-executor framework efficiently mines the memory accesses in the sparse computation, using an access function differentiation approach, to find PSC codelets. It generates vectorized code for the sparse matrix multiplication kernel (SpMV), a kernel with parallel outer loops, and for kernels with carried dependence, specifically the sparse triangular solver (SpTRSV). We demonstrate the performance of the DDF-generated code on a set of 60 large and small matrices (0.05-130M nonzeros). DDF outperforms the highly specialized library MKL with an average speedup of 1.93 and 4.5X for SpMV and SpTRSV, respectively. For the same matrices, DDF outperforms the state-of-the-art inspector-executor framework Sympiler [1] for the SpTRSV kernel by up to 11X and the work by Augustine et. al [2] for the SpMV kernel by up to 12X.