2016/05/15 by Oren Segal, Segal, Oren, Nasibeh Nasiri +3
Computer Science · #Advanced Data Storage Technologies #Distributed and Parallel Computing Systems #FOS: Computer and information sciences #Graph Theory and Algorithms #Hardware Architecture (cs.AR) #Parallel Computing and Optimization Techniques
paper · pdf · doi:10.48550/arxiv.1605.04582
openalex publication_date 2016/05/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Heterogeneous computing can potentially offer significant performance and\nperformance per watt improvements over homogeneous computing, but the question\n"what is the ideal mapping of algorithms to architectures?" remains an open\none. In the past couple of years new types of computing devices such as FPGAs\nhave come into general computing use. In this work we attempt to add to the\nbody of scientific knowledge by comparing Kernel performance and performance\nper watt of seven key algorithms according to Berkley's dwarf taxonomy. We do\nso using the Rodinia benchmark suite on three different high-end hardware\narchitecture representatives from the CPU, GPU and FPGA families. We find\nresults that support some distinct mappings between the architecture and\nperformance per watt. Perhaps the most interesting finding is that, for our\nspecific hardware representatives, FPGAs should be considered as alternatives\nto GPUs and CPUs in several key algorithms: N-body simulations, dense linear\nalgebra and structured grid.\n