High Performance Information Retrieval and MEMS CADon优秀课件教学.pptVIP

  • 0
  • 0
  • 约1.81万字
  • 约 76页
  • 2022-06-08 发布于山东
  • 举报

High Performance Information Retrieval and MEMS CADon优秀课件教学.ppt

Background on Test Matrices Sparse Matrix Benchmark Suite (1/3) # Matrix Name Problem Domain Dimension No. Non-zeros 1 dense Dense matrix 1,000 1.00 M 2 raefsky3 Fluid structure interaction 21,200 1.49 M 3 inaccura Accuracy problem 16,146 1.02 M 4 bcsstk35* Stiff matrix automobile frame 30,237 1.45 M 5 venkat01 Flow simulation 62,424 1.72 M 6 crystk02* FEM crystal free-vibration 13,965 969 k 7 crystk03* FEM crystal free-vibration 24,696 1.75 M 8 nasasrb* Shuttle rocket booster 54,870 2.68 M 9 3dtube* 3-D pressure tube 45,330 3.21 M 10 ct20stif* CT20 engine block 52,329 2.70 M 11 bai Airfoil eigenvalue calculation 23,560 484 k 12 raefsky4 Buckling problem 19,779 1.33 M 13 ex11 3-D steady flow problem 16,214 1.10 M 14 rdist1 Chemical process simulation 4,134 94.4 k 15 vavasis3 2-D PDE problem 41,092 1.68 M Note: * indicates a symmetric matrix. Results on Sun Ultra 1/170 Speedups on SPMV from Sparsity on Sun Ultra 1/170 – 1 RHS Speedups on SPMV from Sparsity on Sun Ultra 1/170 – 9 RHS Preliminary Results on P4 using icc and gcc Speedup of SPMV from Sparsity on P4/icc-5.0.1 Performance of SPMV from Sparsity on P4/icc-5.0.1 Fill for SPMV from Sparsity on P4/icc-5.0.1 Possible Improvements Doesn’t work as well as on Sun Ultra 1/170; Why? Current heuristic to determine best r x c block biased to diagonal of performance plot Didn’t matter on Sun, does on P4 and Itanium since performance so “nondiagonally dominant” Sparsity reg blocking results on P4 for FEM/fluids matrices Matrix #2 (150 Mflops to 400 Mflops) Matrix #5 (50 Mflops to 350 Mflops) Sparsity cache blocking results on P4 for LSI Symmetric Sparse Matrix-Vector Multiply on P4 (vs na?ve full = 1) Sparse Triangular Solve (Matlab’s colmmd ordering) on P4 AT*A on P4 (Accesses A only once) Preliminary Results on Itanium using ecc Speedup of SPMV from Sparsity on Itanium/ecc-5.0.1 Raw Performance of SPMV from Sparsity on Itanium Fill for SPMV from Sparsity on Itanium Possible Improvements Current heuristic to determ

文档评论(0)

1亿VIP精品文档

相关文档