Evaluating Performance and Portability of a Core Bioinformatics Kernel on Multiple Vendor GPUs
Parallel Programming Languages and Models
TimeSunday, 14 November 20213:30pm - 3:55pm CST
DescriptionTraditional scientific simulations have dominated the workloads of high-performance computing infrastructures across the world. With recent advancement in data generation capabilities of systems-biology equipment, a rise in bioinformatics workloads has been observed. Bioinformatics applications deploy algorithmic motifs that are unique to this domain and place unique requirements on modern programming environments as well as GPU accelerators. In this paper, we evaluate the performance and portability of a core bioinformatics kernel that performs DNA and protein sequence alignments in several bioinformatics software pipelines. Our study evaluates the ability of multiple vendor GPUs and their best suited programming models to support these algorithms. We use a highly optimized adaptation of sequence alignment kernel and find the most performant and productive way of porting it across multiple vendor GPUs. Methods used in this paper and the insights drawn from those may prove useful for design and optimization of programming models and hardware accelerators.