DNA sequences alignment in multi-GPUs: acceleration and energy payoff

Abstract

BackgroundWe present a performance per watt analysis of CUDAlign 4.0, a parallel strategy to obtain the optimal pairwise alignment of huge DNA sequences in multi-GPU platforms using the exact Smith-Waterman method.ResultsOur study includes acceleration factors, performance, scalability, power efficiency and energy costs. We also quantify the influence of the contents of the compared sequences, identify potential scenarios for energy savings on speculative executions, and calculate performance and energy usage differences among distinct GPU generations and models. For a sequence alignment on chromosome-wide scale (around 2 Petacells), we are able to reduce execution times from 9.5 h on a Kepler GPU to just 2.5 h on a Pascal counterpart, with energy costs cut by 60%.ConclusionsWe find GPUs to be an order of magnitude ahead in performance per watt compared to Xeon Phis. Finally, versus typical low-power devices like FPGAs, GPUs keep similar GFLOPS/w ratios in 2017 on a five times faster execution.

Hardware Acceleration of the STRIKE String Kernel Algorithm for Estimating Protein to Protein Interactions

“…According to [25], protein-protein interactions (PPIs) can modify the properties of enzymes, play a role in substrate channeling, inactivate a protein, or change the specificity of a protein. For these reasons, PPI assists in the identification of drug targets.…”

“…For these reasons, PPI assists in the identification of drug targets. Many databases support PPI work such as BIND, BioGRID, and MIPS [25]. Prior methods for identifying PPIs are both time-consuming and expensive, including domain knowledge methods, such as AM [1], MLE [2], MSSC [3] and DRF [4], and methods which focus on short polypeptide sequences such as PIPE [5]).…”

“…Other bioinformatics applications on GPU include BLAST [24]. A GPU-based implementation of the SW algorithm [25] on nVIDIA Pascal GPU expedited its execution time by over 300% over an older implementation on nVIDIA Kepler. Hence, bioinformatics algorithms for pairwise similarity can greatly benefit from acceleration by customized hardware, multicore and manycore processors, GPUs, and FPGAs.…”

Accelerating Binary String Comparisons with a Scalable, Streaming-Based System Architecture Based on FPGAs

“…The last approach is the usage of FPGA-accelerators which, although they are not as fast as systems based on high-performance GPUs, tend to be more energy-efficient [22][23][24]. However, these systems have started to reach their limits in terms of performance, therefore, current research focuses on heterogeneous and cluster-based systems, e.g., multi-GPU clusters [25] or CPU-GPU co-systems [26].…”

“…As a follow-up from the last two implementation variants, a mixed design based on lookup tables and DSP blocks was developed in [32]. The focus of this work was to realize population counts for relatively large signatures widths between 2 10 and 2 25 -bit. The best results were achieved with a signature length of 2 20 -bit, with one computation requiring 22 544 µs.…”