FPGA Acceleration of the Phylogenetic Parsimony Kernel?

Abstract

Abstract-The phylogenetic parsimony function is a popular, discrete criterion for reconstructing evolutionary trees based on molecular sequence data. Parsimony strives to find the phylogenetic tree that explains the evolutionary history of organisms by the least number of mutations. Because parsimony is a discrete function, it should fit well to FPGAs. We present a versatile FPGA implementation of the parsimony function and compare its performance to a highly optimized SSE3-and AVX-vectorized software implementation. We find that, because of a particular constellation in our lab, the speedups that can be achieved by using an FPGA, are substantially less impressive, than usually reported in papers on FPGA acceleration of bioinformatics kernels. We conclude that, a competitive spirit between SW and HW application developers can contribute toward obtaining more objective performance comparisons.

Accelerating the phylogenetic parsimony function on heterogeneous systems

“…This subsection evaluates the proposed heterogeneous designs by making comparisons with other state‐of‐the‐art parallel approaches. For this purpose, we conduct comparisons of parallel performance with Parsimonator, which is considered the fastest open‐source implementation of the phylogenetic parsimony function. The CPU‐based release of Parsimonator takes advantage of OpenMP and SSE3/AVX SIMD instructions to accelerate the evaluation of phylogenetic topologies under parsimony.…”

“…Regarding the hardware acceleration of the phylogenetic parsimony function, the most relevant approach was due to Alachiotis and Stamatakis, who reported OpenMP, SSE3/AVX vectorization schemes, and FPGA implementations for Parsimonator. In the research presented in this paper, we aim to go a step further in the study of parallel solutions to perform parsimony computations.…”

Computer science and parsimony: a reappraisal, with discussion of methods for poorly structured datasets

“…At the very least, the flexibility of a program such as TNT should be used to make it run with options that are most appropriate for the dataset at handinstead of selecting the least appropriate, as the majority of the papers discussed here seem to do. Alachiotis and Stamatakis (2011) found similar problems with Kasap and Benkrid's (2010) earlier claim of acceleration of parsimony calculations by hundreds or thousands of times: the virtues of their FPGA implementation had been grossly overestimated due to improper comparisons -Kasap and Benkrid compared their method against a full optimization with PAUP*, but tree searches do not use full optimizations and other programs are faster than PAUP*. Block and Maruyama (2013) did compare FPGA against TNT, and they concluded that their FPGA implementation (massively parallel, but using a full down-pass per rearrangement) was slower than TNT.…”

Partial Reconfiguration of Flux Limiter Functions in MUSCL Scheme Using FPGA

“…Recently, reconfigurable systems using FPGAs have been utilized for acceleration of specific applications including bio-informatics, digital image processing, finance and others [5]- [7]. Even though the early reconfigurable systems did not focus on large scale numerical scientific application, the use of FPGAs for such areas has been growing remarkably because of the rapid performance improvement of modern FPGAs with a large number of configurable logic blocks, memory blocks and embedded multipliers.…”