Understanding Performance Portability of Bioinformatics Applications in SYCL on an NVIDIA GPU

Our goal is to have a better understanding of performance portability of SYCL kernels on a GPU. Toward this goal, we migrate representative kernels in bioinformatics applications from CUDA to SYCL, evaluate their performance on an NVIDIA GPU, and explain the performance gaps through performance profiling and analyses. We hope that the findings provide valuable feedback to the development of the SYCL ecosystem.

I. Introduction

Since graphics-processing unit (GPU) computing platforms are different in the details of vendors’ hardware architectures and software stacks [1], [2], [3], vendor-specific programming libraries and languages have been addressing the differences. For example, CUDA [4] is an advanced programming model widely used for NVIDIA GPUs. However, commonalities among these programming models exist and several portable programming approaches allow for writing code that supports multiple target platforms [5]. In this study, we focus on SYCL, an evolving programming model that facilitates portability across vendors’ devices [6].