High Performance Biological Pairwise Sequence Alignment: FPGA versus GPU versus Cell BE versus GPP

Abstract

This paper explores the pros and cons of reconfigurable computing in the form of FPGAs for high performance efficient computing. In particular, the paper presents the results of a comparative study between three different acceleration technologies, namely, Field Programmable Gate Arrays (FPGAs), Graphics Processor Units (GPUs), and IBM’s Cell Broadband Engine (Cell BE), in the design and implementation of the widely-used Smith-Waterman pairwise sequence alignment algorithm, with general purpose processors as a base reference implementation. Comparison criteria include speed, energy consumption, and purchase and development costs. The study shows that FPGAs largely outperform all other implementation platforms on performance per watt criterion and perform better than all other platforms on performance per dollar criterion, although by a much smaller margin. Cell BE and GPU come second and third, respectively, on both performance per watt and performance per dollar criteria. In general, in order to outperform other technologies on performance per dollar criterion (using currently available hardware and development tools), FPGAs need to achieve at least two orders of magnitude speed-up compared to general-purpose processors and one order of magnitude speed-up compared to domain-specific technologies such as GPUs.

Energy efficiency of sequence alignment tools—Software and hardware perspectives

“…On the contrary, by presenting the performance and energy efficiency of the top-class implementations in this article, we should eliminate this issue and make a more objective comparison. Finally, it is worth noting that according to the study presented in [27], FPGAs outperform all the other platforms on performance per watt criterion, which once again is a promising results in favor of reconfigurable hardware. For this reason, we decided to investigate whether it is possible to further improve the energy efficiency of G-DNA -one of our review implementations, using the FPGA architecture and the RECS R |Box hardware developed within FiPS.…”

“…Another interesting work in this area was published in 2012 [27]. The authors carried out a comparative study between three different acceleration technologies: FPGAs, GPU as well as IBM's Cell BE, and compared them to a traditional CPU.…”

Optimized and Portable FPGA-Based Systolic Cell Architecture for Smith-Waterman-Based DNA Sequence Alignment

“…The proposed cell design implemented in the framework results in 1.75 GCUPS for the affine gap penalty cell and 2.75 for the linear gap penalty cell. This is a 20× and 32× performance improvement over the GPP implementation described in [4]. Table 1 presents the performance improvement.…”

“…Different hardware implementation candidates can be found, such as FPGA, GPU, and cell BE platforms. In [4], a comparison between the FPGA, GPU, and cell BE reconfigurable platforms for aligning biological sequences on the basis of speed, energy consumption, and developmental costs has been presented. Further, it has reported that FPGA outperforms the other two platforms in terms of performance per watt and performance per dollar spent.…”

Reconfigurable acceleration of genetic sequence alignment: A survey of two decades of efforts

“…Basically, most of the research efforts such as [21]- [27] utilize systolic array or fine-grain PE architecture to accelerate PSA with affine gap penalty. Experiments show that, when compared to state-of-the-art software implementations, the reconfigurable accelerators can achieve a speedup from around 40× to 246×.…”