This Article 
 Bibliographic References 
 Add to: 
SPAR: A New Architecture for Large Finite Element Computations
April 1995 (vol. 44 no. 4)
pp. 531-545

Abstract—The finite element method is a general and powerful technique for solving partial differential equations. The computationally intensive step of this technique is the solution of a linear system of equations. Very large and very sparse system matrices result from large finite-element applications. The sparsity must be exploited for efficient use of memory and computational components in executing the solution step. In this paper we propose a scheme, called SPAR, for efficiently storing and performing computations on sparse matrices. SPAR consists of an alternate method of representing sparse matrices and an architecture that efficiently executes computations on the proposed data structure. The SPAR architecture has not been built, but we have constructed a register-transfer level simulator and executed the sparse matrix computations used with some large finite element applications. The simulation results demonstrate a 95% utilization of the floating-point units for some 3D applications. SPAR achieves high utilization of memory, memory bandwidth, and floating-point units when executing sparse matrix computations.

[1] A. Austin,“The PAFEC finite element analysis system,” Finite Element Systems: A Handbook, C.A. Brebbia, ed., Springer-Verlag, New York, 1985.
[2] C. Aykanat,F. Ozguner,S. Martin,, and S.M. Doraivelu,“Parallelization of a finite element application program on a hypercube multiprocessor,” Proc. Second Conf. Hypercube Multiprocessors, SIAM, 1987, pp. 662-673.
[3] K. Bathe,Finite Element Procedure in Engineering Analysis, Prentice Hall, New Jersey, 1982.
[4] R. Chandra,“Conjugate gradient methods for partial differential equations,” Ph.D. Dissertation, Dept. Computer Science, Yale Univ., New Haven, Conn., 1978.
[5] Cray Y-MP Computer Systems Reference Manual, SQ-0083 C, Cray Research Inc., 1985.
[6] Directory of Applications Software for Cray Supercomputers, Cray Research Inc., Jan. 1989.
[7] I.S. Duff,A.M. Erisman,, and J.K. Reid,Direct Methods for Sparse Matrices, Clarendon Press, Oxford, 1986.
[8] C. Farhat,E. Wilson,, and G. Powell,“Solution of finite element systems on concurrent processing computers,” Engineering with Computers, vol. 2, 1987, pp. 157-165.
[9] C. Farhat, and M. Lesoinne,“Automatic partitioning of unstructured meshes for the parallel solution of problems in computational mechanics,” Int’l J. Numerical Methods in Eng., vol. 36, 1993, pp. 745-764.
[10] R.M. Ferencz,“Element-by-element preconditioning techniques for large-scale, vectorized finite element analysis in nonlinear solid and structural mechanics,” Ph.D. Dissertation, Dept. Mechanical Engineering, Stanford Univ., Palo Alto, Calif., 1989.
[11] A. George,“Nested dissection of a regular finite-element mesh,” SIAM J. Numerical Analysis, vol. 10, 1973, pp. 345-363.
[12] A. George,“Numerical experiments using dissection method to solve n by ngrid problems,” SIAM J. Numerical Analysis, vol. 14, Apr. 1977, pp. 161-179.
[13] A. George and J. W.-H. Liu,Computer Solution of Large Sparse Positive Difinite Systems. Englewood Cliffs, NJ: Prentice-Hall, 1981.
[14] D. Goehlich,R.E. Fulton,, and L. Komzsik,“Program designs and data structures for vectorized finite element computations,” Proc. Fourth Int’l Conf. Supercomputing, 1989, pp. 434-438.
[15] G.H. Golub and C.F. Van Loan,Matrix Calculations, The Johns Hopkins Univ. Press, Baltimore, 1983.
[16] L.J. Hayes and P. Devloo,“A vectorized version of a sparse matrix-vector multiply,” Int’l J. Numerical Methods in Eng., vol. 23, 1986, pp. 1043-1056.
[17] C.H. Hamilton and R.S. Gallagher,“PATRAN - The computational laboratory,” Finite Element Systems: A Handbook, C.A. Brebbia, ed., Springer-Verlag, New York, 1985.
[18] S.W. Hammond and K. Law,“Architecture and operation of a systolic engine for finite element computations,” J. Computers and Structures, vol. 30, no. 1/2, Oct. 1988, pp. 365-374.
[19] A.J. Hoffman,M.S. Martin,, and D.J. Rose,“Complexity bounds for regular finite difference and finite element grids,” SIAM J. Numerical Analysis, vol. 10, 1973, pp. 364-369.
[20] J.L. Hennessy and D.A. Patterson, Computer Architecture: A Quantitative Approach, Morgan Kaufmann, San Mateo, Calif., 1990.
[21] M. Heroux, Personal communications, numerical analyst, MathematicalSoftware Group, Cray Research, Eagan, Minn.
[22] M.R. Hestenes and E. Stiefel,“Method of conjugate gradients for solving linear systems,” J. Research of the Nat’l Bureau of Standards, vol. 49, 1952, pp. 409-436.
[23] R.G. Hintz and D.P. Tate,“Control Data STAR-100 processor design,” COMPCON, Sept. 1972, pp. 1-4.
[24] T.J. Hughes,The Finite Element Method: Linear Static and Dynamic Finite Element Analysis, Prentice Hall, New Jersey, 1987.
[25] H.F. Jordan,“A special purpose architecture for finite element analysis,” Proc. 1978 Int’l Conf. Parallel Processing, Aug. 1978, pp. 263-266.
[26] H.F. Jordan and P.L. Sawyer,“A multi-processor system for finite element structural analysis,” Computers and Structures, vol. 10, 1979, pp. 21-29.
[27] D.R. Kincaid,J.R. Respess,D.M. Young,, and R.G. Grimes,“ITPACK 2C: A FORTRAN package for solving large sparse linear systems by adaptive accelerated iterative methods,” ACM Trans. Mathematical Software, vol. 8, 1982, pp. 302-322.
[28] P.M. Kogge,The Architecture of Pipelined Computers, McGraw-Hill, New York, N.Y., 1981.
[29] K.H. Law,“Systolic schemas for finite element methods,” R-82-139, Carnegie Mellon Univ., July 1982.
[30] D.D. Loendorf,“Development and use of an asynchronous MIMD computer for finite element analysis,” Algorithmically Specialized Parallel Computers, Academic Press, 1985.
[31] N. Lincoln,“Technology and design tradeoffs in the creation of modern supercomputers,” IEEE Trans. Computers, vol. 31, 1982, pp. 349-362.
[32] J.W. Liu,“The minimum degree ordering with constraints,” SIAM J. Sci. Stat. Comp., vol. 10, Nov. 1989, pp. 1136-1145.
[33] Lockheed Palo Alto Research Laboratories, Computational Mechanics Section.
[34] N.K. Madsen,G.H. Rodrigue,, and J.I. Karush,“Matrix multiplication by diagonals on a vector/parallel processor,” Information Processing Letters, vol. 5, no. 2, 1976, pp. 41-45.
[35] J.G. Malone,“High performance using a hypercube architecture for parallel nonlinear dynamic finite element analysis,” Proc. Fourth Int’l Conf. Supercomputing, 1989, pp. 434-438.
[36] R. Melhem,“On the design of a pipelined/systolic finite element system,” Computers&Structures, vol. 20, no. 1-3, 1985, pp. 67-75.
[37] R.G. Melhem,“Determination of stripe structures for finite element matrices,” SIAM J. Numerical Analysis, vol. 24, no. 6, Dec. 1987, pp. 1419-1433.
[38] R.G. Melhem,“Solution of linear systems with striped sparse matrices,” Parallel Computing, vol. 6, no. 2, Feb. 1988, pp. 165-184.
[39] R. Melhem,“A systolic accelerator for the iterative solution of sparse linear system,” IEEE Trans. Computers, vol. 38, no. 11, Nov. 1989, pp. 1591-1595.
[40] B. Nour-Omid,A. Raefsky,, and G. Lyzenga,“Solving finite element equations on concurrent computers,” Parallel Computations and Their Impact on Mechanics, A.K. Noor, ed., Am. Soc. Mechanical Engineers, New York, 1987.
[41] B. Nour-Omid and K.C. Park,“Solving structural mechanics problems on the Caltech hypercube machine,” Computer Methods in Applied Mechanics and Eng., vol. 61, 1987, pp. 161-176.
[42] J.M. Ortega, Introduction to Parallel and Vector Solution of Linear Systems, Plenum Press, New York, 1988.
[43] G.V. Paolini,G. Radicati,, and D. Brozolo,“Data structures for vectorizing the conjugate gradient algorithm for general sparsity patterns,” BIT, vol. 29, 1989, pp. 703-718.
[44] J.L. Hennessy and D.A. Patterson, Computer Architecture: A Quantitative Approach, Morgan Kaufmann, San Mateo, Calif., 1990.
[45] J.N. Reddy,An Introduction to the Finite Element Method, McGraw Hill, New York, 1984.
[46] D. Siewiorek, G. Bell, and A. Newell, Computer Structures: Principles and Examples.New York: McGraw-Hill, 1981.
[47] H.D. Simon,“Partitioning of unstructured problems for parallel processing,” Computing Systems in Eng., vol. 2, 1991, pp. 135-148.
[48] O. Storaasli,J. Ransom,, and R. Fulton,“Structural dynamic analysis on a parallel computer: The finite element machine,” Computers and Structures, vol. 26, no. 4, 1987, pp. 551-559.
[49] G. Strang and G.J. Fix,An Analysis of the Finite Element Method, Prentice Hall, New Jersey, 1973.
[50] V.E. Taylor,“Application-specific architectures for large finite-element applications,” Ph.D. Dissertation, Dept. Electrical Engineering and ComputerSciences, Univ. of California at Berkeley, Berkeley, Calif., 1991.
[51] V.E. Taylor,B. Nour-Omid,, and D.G. Messerschmitt,“The effects of communication overhead on the speedup of parallel 3D finite element applications,” Sixth Int’l Parallel Processing Symp., 1992, pp. 531-536.
[52] V.E. Taylor, “Sparse Matrix Computations: Implications for Cache Designs,” Proc. Supercomputing '92, pp. 598-607, 1992.
[53] A. Wolfe,M. Breternitz Jr.,C. Stephens,A.L. Ting,D.B. Kirk,R.P. Bianchini Jr.,, and J.P. Shen,“The White Dwarf: A high-performance application-specific processor,” IEEE Int’l Conf. Computer Architecture, 1988, pp. 212-222.

Index Terms:
Computer architecture, compact data structures, finite element method, resource utilization, sparse matrices.
Abhiram Ranade, Valerie E. Taylor, David G. Messerschmitt, "SPAR: A New Architecture for Large Finite Element Computations," IEEE Transactions on Computers, vol. 44, no. 4, pp. 531-545, April 1995, doi:10.1109/12.376168
Usage of this product signifies your acceptance of the Terms of Use.