Managing the FPGA memory wall: Custom computing or vector processing?
Authors:
- 1 September 2013
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 1946147X,p. 1-6
Abstract
Managing the memory wall is critical for massively parallel FPGA applications where data-sets are large and external memory must be used. We demonstrate that a soft vector processor can efficiently stream data from external memory whilst running computation in parallel. A non-trivial neural computation case study illustrates that multi-core vector processing coupled with careful layout of data structures performs similarly to an elaborate full-custom memory controller and execution pipeline. The vector processing version was far simpler to code so we encourage others to consider vector machines before contemplating a full-custom architecture on FPGA.Keywords
This publication has 8 references indexed in Scilit:
- A Large-Scale Model of the Functioning BrainScience, 2012
- Bluehive - A field-programable custom computing machine for extreme-scale real-time neural network simulation2012
- VENICE: A Compact Vector Processor for FPGA Applications2012
- VEGAS2011
- Vector Processing as a Soft Processor AcceleratorACM Transactions on Reconfigurable Technology and Systems, 2009
- VESPA2008
- Reflections on the memory wall2004
- Simple model of spiking neuronsIEEE Transactions on Neural Networks, 2003