CUDASW++ 3.0: accelerating Smith-Waterman protein database search by coupling CPU and GPU SIMD instructions

Abstract

BackgroundThe maximal sensitivity for local alignments makes the Smith-Waterman algorithm a popular choice for protein sequence database search based on pairwise alignment. However, the algorithm is compute-intensive due to a quadratic time complexity. Corresponding runtimes are further compounded by the rapid growth of sequence databases.ResultsWe present CUDASW++ 3.0, a fast Smith-Waterman protein database search algorithm, which couples CPU and GPU SIMD instructions and carries out concurrent CPU and GPU computations. For the CPU computation, this algorithm employs SSE-based vector execution units as accelerators. For the GPU computation, we have investigated for the first time a GPU SIMD parallelization, which employs CUDA PTX SIMD video instructions to gain more data parallelism beyond the SIMT execution model. Moreover, sequence alignment workloads are automatically distributed over CPUs and GPUs based on their respective compute capabilities. Evaluation on the Swiss-Prot database shows that CUDASW++ 3.0 gains a performance improvement over CUDASW++ 2.0 up to 2.9 and 3.2, with a maximum performance of 119.0 and 185.6 GCUPS, on a single-GPU GeForce GTX 680 and a dual-GPU GeForce GTX 690 graphics card, respectively. In addition, our algorithm has demonstrated significant speedups over other top-performing tools: SWIPE and BLAST+.ConclusionsCUDASW++ 3.0 is written in CUDA C++ and PTX assembly languages, targeting GPUs based on the Kepler architecture. This algorithm obtains significant speedups over its predecessor: CUDASW++ 2.0, by benefiting from the use of CPU and GPU SIMD instructions as well as the concurrent execution on CPUs and GPUs. The source code and the simulated data are available at http://cudasw.sourceforge.net.

Aplicación del modelo de programación CUDA en la simulación de la evolución de secuencias genéticas

“…CUDA (Compute Unified Device Architecture) es una plataforma de computación paralela y un modelo de programación inventado por NVIDIA ("CUDA Zone | NVIDIA Developer," 2011; Nvidia, 2011;NVIDIA, 2015) permite aumentos impresionantes en el rendimiento del computador al aprovechar las unidades de procesamiento gráfico (GPU) alojadas en la tarjeta gráfica. Actualmente se reportan varias aplicaciones de esta tecnología en el área de la Bioinformática: en el análisis de genes en microarreglos y alineación de secuencias (Yongchao Liu, Schmidt, & Maskell, 2012), simulación de sistemas biológicos y búsqueda de secuencias en bases de datos (Y Liu, Wirawan, & Schmidt, 2013), entre otras aplicaciones disponibles en el sitio para desarrolladores de NVIDIA ("CUDA Zone | NVIDIA Developer," 2011), (Hwu, 2012), (Schatz, Trapnell, Delcher, & Varshney, 2007), (Manavski & Valle, 2008).…”

An Efficient GPUAccelerated Implementation of Genomic Short Read Mapping with BWAMEM

“…ted implementations of this algorithm exist (e.g., [8], [11]). However, all these implementations perform one complete sequence matching per compute thread, making such an implementation unsuitable for direct application onto BWA-MEM, as it requires batching and sorting of larger groups of work.…”

Remote GPU Virtualization: Is It Useful?

“…Finally, virtual machines were configured to have 4 cores and 12 GB of RAM memory. The applications used in this analysis are LAMMPS [2], CUDA-MEME [16], CUDASW++ [9], and GPU-BLAST [27], being all of them listed in the NVIDIA GPU-Accelerated Applications Catalog [13]. Figure 7 shows the performance of these four applications when executed in the following scenarios:…”