
326 Designing Scientific Applications on GPUs
offset2
offset1
offset0
offset3
right−copy
Initial sparse matrix
left−copy
right−copy
Global sparse matrix
left−copy
Node 0
Node 1
Node 3
Node 2
FIGURE 13.6. Parallel generation of a large sparse matrix by four computing
nodes.
Matrix Type Matrix Name # Nonzeros Bandwidth
Symmetric
2cubes sphere 413, 703, 602 198, 836
ecology2 124, 948, 019 2, 002
finan512 278, 175, 945 123, 900
G3 circuit 125, 262, 292 1, 891, 887
shallow water2 100, 235, 292 62, 806
thermal2 175, 300, 284 2, 421, 285
Nonsymmetric
cage13 435, 770, 480 352, 566
crashbasis 409, 291, 236 200, 203
FEM 3D thermal2 595, 266, 787 206, 029
language 76, 912, 824 398