Application of Memory-Based Computing
- 754 Downloads
We propose a hardware architecture to utilize a dense 2-D memory array for the purpose of reconfigurable computing. The main idea is to map multiple multi-input multi-output LUTs to the embedded memory array at each compute block and evaluate them over multiple clock cycles. The proposed framework is therefore a spatio-temporal framework unlike the conventional hardware reconfigurable frameworks which are fully spatial. In the proposed hardware architecture, multiple computing elements communicate with each other over a time-multiplexed programmable interconnect.
Along with the hardware architecture we outline an effective software framework which maps an input application to the proposed hardware reconfigurable framework. The software flow outlined partitions the input application and maps it to multiple compute blocks, and finally places and routes the mapped design.
We describe the application domains of the proposed memory based computing model. We identify that the model can be applied to realize a stand-alone reconfigurable framework for mapping random logic as well as it can be used for hardware acceleration of algorithmic tasks.
KeywordsComputer-based Memory Input Application Memory Array Programmable Interconnect Random Logic
- 1.A. Dehon, “DPGA Utilization and Application”, in Intl. Symp. on FPGAs, 1996Google Scholar
- 2.E.S. Chung, P.A. Milder, J.C. Hoe, K. Mai, “Single-Chip Heterogeneous Computing: Does the Future Include Custom Logic, FPGAs, and GPGPUs? in Intl. Symp. on Microarchitecture, 2010Google Scholar
- 3.M. Bushnell, V. Agarwal, “Essentials of Electronic Testing for Digital, Memory and Mixed-Signal VLSI Circuits” (Springer, Heidelberg, 2000)Google Scholar
- 5.S. Che, J. Li, J.W. Sheaffer, K. Skadron, J. Lach, “Accelerating Compute-Intensive Applications with GPUs and FPGAs”, in Symp. on Application Specific Processor, 2008Google Scholar
- 6.R. Tessier, W. Burleson, “Reconfigurable computing for digital signal processing: A survey”. J. VLSI Signal Process. Syst. 28(1/2), (2001)Google Scholar
- 7.S. Paul, S. Bhunia, “Dynamic transfer of computation to processor cache for yield and reliability improvement”. IEEE Trans. Very Large Scale Integrat. Syst., 1368–1379 (2011)Google Scholar
- 8.J. Dido et al., “A Flexible Floating Point Format for Optimizing Data-Paths and Operators FPGA Based DSPs”, in Intl. Symp. on FPGAs, 2002Google Scholar