Power Analysis of a General Convolution Algorithm Mapped on a Linear Processor Array

We explore the energy dissipation of the Linear Processor Array (LPA) as a function of the number of available resources (Processor Units P) within the array. This number P is an important parameter, as it reflects performance, relates parallel processing to energy dissipation, and influences the scaling of the various parts of the LPA architecture (memory, address generator, communication network).To make a comparison of the different design variants for a fixed datawidth possible, we propose a high-level energy dissipation model of the processor, which is based on a detailed analysis of a general convolution algorithm.It is shown that the energy dissipation of the LPA can roughly be described by the relationship Etotal ∼ N/P with N presenting the datawidth in pixels. This relationship is derived from two observations: first, the largest contribution to Etotal is formed by the energy dissipated by the memories, and second, in our model of the LPA, the datawidth of the memories corresponds with the number of pixels N to be processed, which results in an increase of the access rate when P decreases.Furthermore, we have shown that the energy dissipation caused by communication within the LPA, increases with increasing number of resources: the trade-off between communication versus computation in parallel computing. This turns out to be negligible in the total energy dissipation, and we therefore conclude, that the optimum solution is found, when a full number of resources is applied within the LPA.

[1]  Christer Svensson,et al.  An addressable 256×256 photodiode image sensor array with an 8-bit digital output , 1993, ESSCIRC '92: Eighteenth European Solid-State Circuits conference.

[2]  Wim F. J. Verhaegh,et al.  PHIDEO: High-level synthesis for high throughput applications , 1995, J. VLSI Signal Process..

[3]  Christer Svensson,et al.  PASIC: A processor-A/D converter-sensor integrated circuit , 1990, IEEE International Symposium on Circuits and Systems.

[4]  Pieter P. Jonker.  Why linear arrays are better image processors , 1994, Proceedings of the 12th IAPR International Conference on Pattern Recognition, Vol. 2 - Conference B: Computer Vision & Image Processing. (Cat. No.94CH3440-5).

[5]  Robert H. Halstead,et al.  Computation structures , 1990, MIT electrical engineering and computer science series.

[6]  Kees G. W. Goossens,et al.  The Petrol approach to high-level power estimation , 1998, Proceedings. 1998 International Symposium on Low Power Electronics and Design (IEEE Cat. No.98TH8379).

[7]  E. R. Komen,et al.  Low-level image processing architectures: Compared for some non-linear recursive neighbourhood operations , 1990 .

[8]  S. Kung,et al.  VLSI Array processors , 1985, IEEE ASSP Magazine.

[9]  Anantha P. Chandrakasan,et al.  Low Power Digital CMOS Design , 1995 .

[10]  Gealow,et al.  A Pixel-parallel Image Processor Using Logic Pitch-matched To Dynamic Memory , 1997, Symposium 1997 on VLSI Circuits.

[11]  Massoud Pedram,et al.  Low power design methodologies , 1996 .

[12]  Massoud Pedram,et al.  Power Aware Design Methodologies , 2002 .

[13]  Paul Wielage,et al.  Xetal: a low-power high-performance smart camera processor , 2001, ISCAS 2001. The 2001 IEEE International Symposium on Circuits and Systems (Cat. No.01CH37196).

[14]  Teresa H. Y. Meng,et al.  Low-power DV encoder architecture for digital CMOS camcorder , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[15]  Narayanan Vijaykrishnan,et al.  A linear array processor with dynamic frequency clocking for image processing applications , 1998, IEEE Trans. Circuits Syst. Video Technol..