稀疏对称矩阵的LDLT分解在GPU上的高效实现
|
陈鑫峰,王武
|
An Effective Implementation of LDLT Decomposition of Sparse Symmetric Matrix on GPU
|
Chen Xinfeng,Wang Wu
|
|
表1 基于本文的直接法求解、LDLT分解和求解阶段的时间(ms)
|
Table 1 Runtime (in ms) of LDLT decomposition and solving phases using our direct solver
|
|
matrix | n | nz | nnz | nnz/nz | symbolic | numeric | solve | total | windscreen | 22692 | 752541 | 5545914 | 7.370 | 311.814 | 973.928 | 534.656 | 1820.398 | crystk03 | 24696 | 887937 | 9221958 | 10.386 | 371.007 | 693.275 | 88.267 | 1152.549 | bcsstk37 | 25503 | 583240 | 2996850 | 5.138 | 182.664 | 342.704 | 29.465 | 554.833 | bcsstk35 | 30237 | 740200 | 3045171 | 4.114 | 178.499 | 354.06 | 31.053 | 563.612 | t3dh | 79171 | 2215638 | 45191167 | 20.396 | 3311.172 | 4136.49 | 440.249 | 7887.911 | TEM152078 | 152078 | 3305720 | 57409931 | 17.367 | 4423.406 | 4893.649 | 556.43 | 9873.485 | TEM181302 | 181302 | 4010156 | 70510354 | 17.583 | 5555.54 | 6092.029 | 689.041 | 12336.61 | pwtk | 217918 | 5926171 | 47124510 | 7.952 | 2570.829 | 2878.245 | 468.653 | 5917.727 | BenElechi1 | 245874 | 6698185 | 52230259 | 7.798 | 2779.673 | 3193.183 | 176.464 | 6095.320 |
|
|
|