K. A. Gallivan, B. A. Marsolf, and H. A. G. Wijshoff. MCSPARSE: A parallel sparse unsymmetric linear system solver. TR. 1142, CSRD, Univ. of Illinois, Urbana, IL, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... while loops that could not be parallelized by any compiler available to us; two loops are from the PERFECT Benchmarks [3] two loops are from MA28, a sparse non symmetric linear solver [5] and one loop is extracted from MCSPARSE, a parallel version of a non symmetric sparse linear systems solver [6, 7]. Our results are summarized in Table 2. For each method applied to a loop, we give the speedup that was obtained, and, mention whether backups and time stamping were necessary. Whenever necessary, we performed a simple preventive backup of the variables potentially written in the loop. In some ....

K. A. Gallivan, B. A. Marsolf, and H. A. G. Wijshoff. MCSPARSE: A parallel sparse unsymmetric linear system solver. TR. 1142, CSRD, Univ. of Illinois, Urbana, IL, 1991.

....input matrix into a bordered block upper triangular structure. This structure can be used to exploit large, medium and fine grained parallelism during the subsequent factorization and solve phases. The initial implementation of MCSPARSE was specifically targeted for the Cedar architecture [7, 8, 14, 16]. The Cedar is a cluster based multivector machine containing a hybrid memory system. Later, a shared memory version of MCSPARSE is developed and implemented for the Cray Y MP. Also, the preprocessing phase H is further optimized and parallelized [11, 12] In this paper, we specifically ....

....offered by the target machines Each of the designs for the three machine models propose several different mappings. In this section, we concentrate on comparing the most efficient mappings for the Cedar, Cray Y MP and iPSC 860. A detailed discussion of the other mappings can be found in [7], 11] and earlier sections in this paper respectively. The Cedar architecture consists of four clusters, which are loosely coupled through a global memory. Each cluser comprises 8 vector processors, tightly coupled through a shared memory. The diagonal blocks are distributed among the multiple ....

K. A. Gallivan, B. A. Marsolf and H. A. G. Wijshoff, MCSPARSE : A Parallel Sparse Unsymmetric Linear System Solver, Tech. Rep. CSRD Report No. 1142, Center for Supercomputing Research and Development, University of Illinois, Urbana, IL, 1991.

.... not be parallelized by any compiler available to us; two loops are from the PERFECT Benchmarks [BCK 89] two loops are from MA28, a sparse non symmetric linear solver [Duf77] and one loop is extracted from MCSPARSE, a parallel version of a non symmetric sparse linear systems solver [GMW89, GMW91] Our results are summarized in Table 8.9. For each method applied to a loop, we give the speedup that was obtained, and, mention whether backups and time stamping were necessary. Whenever necessary, we performed a simple preventive backup of the variables potentially written in the loop. In some ....

K. A. Gallivan, B. A. Marsolf, and H. A. G. Wijshoff. MCSPARSE: A parallel sparse unsymmetric linear system solver. Technical Report 1142, Center for Supercomputing Research and Development, University of Illinois, Urbana, IL, 1991.

.... WHILE loops that could not be parallelized by any compiler available to us; two loops are from the PERFECT Benchmarks [3] two loops are from MA28, a sparse UNsymmetric linear solver [6] and one loop is extracted from MCSPARSE, a parallel version of a non symmetric sparse linear systems solver [7, 8]. Our results are summarized in Table 2. For each method applied to a loop, we give the speedup that was obtained, and, mention whether backups and time stamping were necessary. Whenever necessary, we performed a simple preventive backup of the variables potentially written in the loop. In some ....

K. A. Gallivan, B. A. Marsolf, and H. A. G. Wijshoff. MCSPARSE: A parallel sparse unsymmetric linear system solver. Technical Report CSRD Report No. 1142, Center for Supercomputing Research and Development, University of Illinois, Urbana, IL, 1991.

....in parallel. Similarly, one can see that the off diagonal blocks can also be updated concurrently. Of course, each off diagonal block can only be updated until after its corresponding diagonal block has been factorised. How global stability and sparsity can be maintained is described in detail in [3]. The elimination of a particular border row can be done independently of the elimination of the other border rows. Therefore, the elimination of the border can be done in parallel over its comprising border rows. A restriction within the elimination of a specific border row is that its ....

....blocks must be applied in the order in which they appear in the reordered matrix. Of course, each diagonal block can only be applied until after its factorisation and the update of its off diagonal block have been completed. Again, specifics concerning stability and sparsity can be found in [3]. save old flow status in Bi calculate time step in Bi apply navier stokes in Bi apply euler in Bi O advance flow status Bi s1 read input apply local boundary conditions in Bi ld1 gd1 gd2 gd3 apply g.b.c. on faces in Bi apply g.b.c. on edges in Bi gd4 gd5 apply local boundary conditions in Bi ....

K.A. Gallivan, B.A. Marsolf and H.A.G. Wijshoff, MCSPARSE: A parallel sparse unsymmetric linear system solver, Tech. Rep. CSRD Report No. 1142, Center for SupercomputingResearch and Development, University of Illinois, Urbana, IL, 1991

....problems 2 and 3 from [20] All matrices are stored in sparse column wise format [23] no symmetry is exploited. The matrices contain real entries only. Some of them are supplied with a right hand side. If there was no right hand side available, a random one was generated. The H0 reordering [8] was applied prior to factorization to make sure that the diagonal did not contain any structural zeros. As can be seen from table 1, these matrices are not large. In practice a direct solver would be used to solve matrices of this size. For efficiency comparison they are large enough as long as ....

K.A. Gallivan, B.A. Marsolf, and H.A.G. Wijshoff. MCSPARSE: A parallel sparse unsymmetric linear system solver. Technical Report 1100, Center for Supercomputing Research and Development, University of Illinois, Urbana, IL, 1990.

No context found.

K.A. Gallivan, B.A. Marsolf, and H.A.G. Wijshoff. Mcsparse: A parallel sparse unsymmetric linear system solver. Technical Report no. 1142, Center for Supercomputing Research and Development, University of Illinios, 1991.

....APPARC (No. 6634) the performance. Both of these approaches can be used as part of an algorithm which exploits multiple levels of parallelism. Tearing techniques have been proposed to expose large grain structure and parallelism by reordering the matrix into a bordered block triangular matrix[14, 20, 31]. This effectively partitions the problem into small subproblems (the diagonal blocks) and then eliminates all connections between the subproblems (the border blocks) Unfortunately, the associated factorization routines are often unable to preserve stability and sparsity without destroying this ....

....P Pq 9 : X X X Xy i C S B a b c d e f g h 2. After enhancement i 1. Before reduction h g f e d c b a B S C i h g f e d c b a B S C 3. After reduction FIG. 5. Enhanced Separator Set Reduction performance results which include the ordering time. The interested reader should consult [20] for details concerning the tuning of the heuristics that produces the data presented below. In this paper, we restrict ourselves to the description of the following parameter settings which were used for each of the phases of H . The choice for each of these parameters is based on the given ....

[Article contains additional citation context not shown here]

K. A. GALLIVAN, B. A. MARSOLF, AND H. A. G. WIJSHOFF, MCSPARSE: A parallel sparse unsymmetric linear system solver, Tech. Report CSRD Report No. 1142, Center for Supercomputing Research and Development, University of Illinois, Urbana, IL, 1991.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC