Chapter 10. Programming Data Parallel FPGA Applications Using the SIMD/Vector Model

Maya B. GokhaleLawrence Livermore National Laboratory

In the Single Instruction Multiple Data (SIMD) model, aggregate operations on arrays and vectors can be mapped to arrays of function units. A single instruction stream is dispatched from a control unit to the function units, which operate in lockstep on the data sequences. Reconfigurable hardware is well suited to perform SIMD (also called vector or data parallel) computation (see Section 5.1.5). Groups of lookup tables (LUTs) can be configured as function units, and the data local to each unit can be stored in distributed memories. This chapter explores parallel processing on reconfigurable computers using the ...

Get Reconfigurable Computing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.