Author : Sambhu P.G 1
Date of Publication :7th April 2016
Abstract: DSP accelerators are hardware modules attached to a processor core externally to enhance the performance and functionality of computationally intensive DSP functions. Domain specific hardware architectures forms ideal acceleration in terms of performance and power, but their inflexible data paths lead to increased silicon complexity. In flexible DSP accelerator functional computational unit (FCU) are incorporated to improve performance, reduce energy consumption and to provide flexible data paths. The architecture exploits carry save (CS) arithmetic to enable fast chaining of additive and multiplicative operations. However, the carry save optimization approaches have limited impact on data flow graph (DFG) dominated by multiplications. These limitations are tackled by exploiting CS to modified booth recoding, which consists of three algorithms, signed-bit Full Adders (FAs) and Half Adders (HAs) as building blocks. To validate the design, code can be developed using VHDL and synthesis & simulation will be done in Xilinx ISE Suite 13.2 & Model Sim SE 6.3f respectively.
Reference :
-
[1]K. Tsoumanis, S. Xydis, C. Efstathiou, N. Moschopoulos, and K. Pekmestzi, “Flexible DSP Accelerator Architecture Exploiting Carry-Save Arithmetic”, IEEE Trans.Circuits Syst. I, Reg. Papers, volume: 61, no.4, January 2015.
[2] K. Tsoumanis, S. Xydis, C. Efstathiou, N. Moschopoulos, and K. Pekmestzi,“An optimized modified booth recoder for efficient design of the add-multiply operator”, IEEE Trans. Circuits Syst. I, Reg. Papers, volume: 61, no.4, April 2014.
[3] M. Stojilovic, D. Novo, L. Saranovac, P. Brisk, and P. Ienne, “Selective flexibility: Creating domain-specific reconfigurable arrays”, IEEE Trans.Comput.-Aided Design Integr. Circuits Syst., volume:32, no.5,May 2013.
[4] G. Ansaloni, P. Bonzini, and L. Pozzi, “EGRA: A coarse grained reconfigurable architectural template”, IEEE Trans. Very Large Scale Integr. (VLSI) Syst., volume: 19, no. 6, June. 2011.
[5] S. Xydis, G. Economakos, D. Soudris, and K. Pekmestzi, “High performance and area efficient flexible DSP datapath synthesis” IEEE Trans.Very Large Scale Integr. (VLSI) Syst., volume: 19, no. 3, March. 2011.
[6] K. Compton and S. Hauck,“Automatic design of reconfigurable domain specific flexible cores”, IEEE Trans. Very Large Scale Integr. (VLSI) Syst.,volume: 16, no.5,May 2008.
[7] M. D. Galanis, G. Theodoridis, S. Tragoudas, and C. E. Goutis, “A high performance data path for synthesizing DSP kernels”, IEEETrans. Comput.- Aided Design Integr. Circuits Syst., volume: 25, no. 6, June. 2006.