CUDASW++4.0: Ultra-fast GPU-based Smith-Waterman Protein Sequence Database Search

Abstract

Background: The maximal sensitivity for local pairwise alignment makes the Smith-Waterman algorithm a popular choice forprotein sequence database search. However, its quadratic time complexity makes it compute-intensive. Unfortunately, current state-of-the-art software tools are not able to leverage the massively parallel processing capabilities of modern GPUs with close-to-peak performance. This motivates the need for more efficient implementations. Results: CUDASW++4.0 is a fast software tool for scanning protein sequence databases with the Smith-Waterman algorithm on CUDA-enabled GPUs. Our approach achieves high efficiency for dynamic programming-based alignment computation by minimizing memory accesses and instructions. We provide both efficient matrixtiling and sequence database partitioning schemes, and exploit next generation floating point arithmetic and novel DPX instructions. This leads to close-to-peak performance on modern GPU generations (Ampere, Ada, Hopper) with throughput rates of up to 1.94 TCUPS, 5.01 TCUPS, 5.71 TCUPS on an A100, L40S, and H100, respectively. Evaluation on the Swiss-Prot, UniRef50, and TrEMBL databases shows that CUDASW++4.0 gains over an order-of-magnitude performance improvements over previous GPU-based approaches (CUDASW++3.0, ADEPT). In addition, our algorithm demonstrates significant speedups over top-performing CPU-based tools (BLASTP, SWIPE, SWIMM2.0), can exploit multi-GPU nodes with linear scaling, and features an impressive energy efficiency of up to 15.7 GCUPS/Watt. Conclusion: CUDASW++4.0 changes the standing of GPUs in protein sequence database search with Smith-Waterman alignment by providing close-to-peak performance on modern GPUs. It is freely available at https://github.com/asbschmidt/CUDASW4.

iPuma: High-Performance Sequence Alignment on the Graphcore IPU

“…Recent GPU implementations, such as AnySeq [35] CUD-ASW++ 4.0 [36] have claimed promising performance charac-teristics, especially when utilizing novel DPX instruction set features available on the NVIDIA Hopper micro-architecture.…”

Estudio de viabilidad de SYCL como modelo de programación unificado para sistemas heterogéneos basados en GPUs en bioinformática

“…El objetivo del alineamiento es comparar dos o más secuencias biológicas para identificar las regiones que comparten una historia evolutiva común. Por esta razón, existen varias aplicaciones destacadas como CUDAlign [31,32,33,34,35], SW# [36], SW#db [37], CUDASW [38,39,40,41], ADEPT [42], entre otras. Sin embargo, debido a que en general se han desarrollado sobre CUDA, la portabilidad a otras arquitecturas se ve gravemente limitada, lo que restringe explotar las ventajas que pueden presentar los diferentes dispositivos actuales o emergentes.…”

“…En 2023, Bertil Schmidt et al presentaron CUDASW++4.0 [41], una herramienta de software avanzada para el escaneo de bases de datos de secuencias proteicas utilizando el algoritmo de SW en GPUs compatibles con CUDA. Esta versión logra una alta eficiencia en el cálculo de alineamientos basadas en programación dinámica, minimizando el acceso a la memoria y las instrucciones ejecutadas.…”