SWIMM 2.0: Enhanced Smith–Waterman on Intel’s Multicore and Manycore Architectures Based on AVX-512 Vector Extensions

Abstract

The well-known Smith-Waterman (SW) algorithm is the most commonly used method for local sequence alignments, but its acceptance is limited by the computational requirements for large protein databases. Although the acceleration of SW has already been studied on many parallel platforms, there are hardly any studies which take advantage of the latest Intel architectures based on AVX-512 vector extensions. This SIMD set is currently supported by Intel's Knights Landing (KNL) accelerator and Intel's Skylake (SKL) general purpose processors. In this paper, we present an SW version that is optimized for both architectures: the renowned SWIMM 2.0. The novelty of this vector instruction set requires the revision of previous programming and optimization techniques. SWIMM 2.0 is based on a massive multi-threading and SIMD exploitation. It is competitive in terms of performance compared with other state-of-the-art implementations, reaching 511 GCUPS on a single KNL node and 734 GCUPS on a server equipped with a dual SKL processor. Moreover, these successful performance rates make SWIMM 2.0 the most efficient energy footprint implementation in this study achieving 2.94 GCUPS/Watts on the SKL processor.

BLVector: Fast BLAST-Like Algorithm for Manycore CPU With Vectorization

“…And, in real world scenarios, it produces the same results than blastp, but faster. Filtering a sequence through the 4-mer method is up to 11.8-fold quicker than calculating its S/W score, so it can be applied not only to Farrar algorithm, but also to any brute-force approach ( Rognes, 2011 ; Rucci et al, 2019 ), as a way to speed it up.…”

“…Other approaches could be used to align many subject sequences with extreme performance. However, they do not include a filtering stage and/or they require specific servers and/or high-end graphics cards, whose cost greatly exceeds the affordable Xeon Phi Coprocessor 31S1P that BLVector may use ( Lan et al, 2017 ; Rucci et al, 2019 ). Anyway, the BLVector source code could be rewritten to get the most out of any brand new hardware vectorization.…”

An Overview of Duplicated Gene Detection Methods: Why the Duplication Mechanism Has to Be Accounted for in Their Choice

“…The most popular ones such as BLAST [ 115 ], Psi-BLAST [ 116 ], and HMMER3 [ 117 ] are heuristic methods. Thus, they might not give the best results, but they drastically save computational time compared to a classical method such as the Smith and Waterman algorithm [ 118 ] even though some implementations have tried to make it faster as PARALIGN [ 119 ] or SWIMM [ 120 ]. In the case of homology, the alignment is generally performed on protein sequence instead of the gene.…”

An Efficient and Rapid Method for Detection of Mutations in Deoxyribonucleic Acid - Sequences

“…In another study carried out in 2019 by [13], a new solution was used, based on massive multithreaded exploitation with a focus on the latest Intel architectures based on Advanced Vector Extensions 512 (AVX-512). The goal is to address the limited acceptance of the Smith-Waterman algorithm by the computational requirements of large protein databases often used for local sequence alignment.…”