Energy-Quality Scaling in Analog Mesh Computers

The recent push for post-Moore computer architectures has introduced a wide variety of application-specific accelerators. One particular accelerator, the resistance network analogue, has been well received due to its ability to efficiently solve partial differential equations by eliminating the iterative stages required by today's numerical solvers. However, in the ago of programmable integrated circuits, the static nature of the resistance network analogue, and other analog mesh computers like it, has relegated it to an academic curiosity. Recent developments in materials, such as the memristor, have made the resistance network analogue viable for inclusion in future heterogeneous computer architectures. However, selection of an appropriate sized mesh to be incorporated into a computer system requires that energy-quality trade-offs are made regarding the problem size and required resolution of the solution. This paper provides an in-depth study of the scaling of analog mesh computer hardware, from the perspective of energy per bit and required resolution, introduces a metric to aid in quantifying analog mesh computers with different parameters, and introduces a method of virtualization which enables an analog mesh computer of a fixed size to approximate the calculations of a larger-sized mesh.

[1]  Chen Tang,et al.  PDE (ODE)-based image processing methods for optical interferometry fringe , 2013, Other Conferences.

[2]  Francisco Tirado,et al.  A special-purpose parallel computer for solving partial differential equations , 1998, Proceedings of the Sixth Euromicro Workshop on Parallel and Distributed Processing - PDP '98 -.

[3]  Sihwan Kim,et al.  A Programmable and Configurable Mixed-Mode FPAA SoC , 2016, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[4]  Vikram K. Narayana,et al.  Towards a Computational Model for Heat Transfer in Electrolytic Cells , 2014 .

[5]  Massimo Alioto,et al.  Energy-quality scalable adaptive VLSI circuits and systems beyond approximate computing , 2017, Design, Automation & Test in Europe Conference & Exhibition (DATE), 2017.

[6]  Costas D. Sarris,et al.  Adaptive Mesh Refinement for Time-Domain Numerical Electromagnetics , 2006, Adaptive Mesh Refinement for Time-Domain Numerical Electromagnetics.

[7]  Tien Mo Shih Numerical Heat Transfer , 1984 .

[8]  Thierry Blu,et al.  How a simple shift can significantly improve the performance of linear interpolation , 2002, Proceedings. International Conference on Image Processing.

[9]  Jianping Zhu Solving partial differential equations on parallel computers , 1994 .

[10]  Ronald L. Rivest,et al.  Introduction to Algorithms, 3rd Edition , 2009 .

[11]  Engin Ipek,et al.  Memristive Boltzmann machine: A hardware accelerator for combinatorial optimization and deep learning , 2017 .

[12]  G Liebmann Resistance-network analogues with unequal meshes or subdivided meshes , 1954 .

[13]  Sabine Himmel,et al.  Partial Differential Equations For Scientists And Engineers , 2016 .

[14]  Tao Zhang,et al.  PRIME: A Novel Processing-in-Memory Architecture for Neural Network Computation in ReRAM-Based Main Memory , 2016, 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA).

[15]  William E. Schiesser The numerical method of lines , 1991 .

[16]  A. Hastings The Art of Analog Layout , 2000 .

[17]  Yu Lin,et al.  Resistors layout for enhancing yield of R-2R DACs , 2002, 2002 IEEE International Symposium on Circuits and Systems. Proceedings (Cat. No.02CH37353).

[18]  E. Lorenz Deterministic nonperiodic flow , 1963 .

[19]  Wolfgang Joppich,et al.  Practical Fourier Analysis for Multigrid Methods , 2004 .

[20]  R. Jacob Baker,et al.  Gain Error Correction for CMOS Image Sensor Using Delta-Sigma Modulation , 2010, 2010 IEEE Workshop on Microelectronics and Electron Devices.

[21]  Sparsh Mittal,et al.  A Survey of Techniques for Approximate Computing , 2016, ACM Comput. Surv..

[22]  M. Berger,et al.  Adaptive mesh refinement for hyperbolic partial differential equations , 1982 .

[23]  Marcus Randall,et al.  Application specific computers for combinatorial optimisation , 1997 .

[24]  A.P.S. Selvadurai,et al.  Fundamentals, Laplace's equation, diffusion equation, wave equation , 2000 .

[25]  Michael R. Foster,et al.  Partial differential equations in fluid dynamics , 2008 .

[26]  Thomas Unterluggauer,et al.  A low-area ASIC implementation of AEGIS128 — A fast authenticated encryption algorithm , 2014, 22nd Austrian Workshop on Microelectronics (Austrochip).

[27]  Robert M. Corless,et al.  A New View of the Computational Complexity of IVP for ODE , 2002, Numerical Algorithms.

[28]  Catherine Graves,et al.  Dot-product engine for neuromorphic computing: Programming 1T1M crossbar to accelerate matrix-vector multiplication , 2016, 2016 53nd ACM/EDAC/IEEE Design Automation Conference (DAC).

[29]  A.P.S. Selvadurai,et al.  Partial differential equations in mechanics , 2000 .

[30]  Eriko Nurvitadhi,et al.  Can FPGAs Beat GPUs in Accelerating Next-Generation Deep Neural Networks? , 2017, FPGA.

[31]  Dean M. Tullsen,et al.  Harnessing ISA diversity: Design of a heterogeneous-ISA chip multiprocessor , 2014, 2014 ACM/IEEE 41st International Symposium on Computer Architecture (ISCA).

[32]  G Liebmann,et al.  Solution of Partial Differential Equations with a Resistance Network Analogue , 1950 .

[33]  T. Sandeep,et al.  A Complete Analysis of Tolerance of Component in Analog VLSI Circuits Using Sensitivity , 2016 .

[34]  G Liebmann,et al.  An improved experimental iteration method for use with resistance-network analogues , 1954 .