. 2013 Sep 20;30(1):117–134. doi: 10.1002/cnm.2593

Table II.

Speedup of the parallel CPU and GPU implementations with respect to the sequential CPU version.

Stage	1_CPU/32_CPU	1_CPU/1_GPU	32_CPU/1_GPU
Jacobian	22.3 ×	44.5 ×	2.0 ×
Residual	20.4 ×	34.8 ×	1.7 ×
PDE solve	3.7 ×	1.0 ×	0.3 ×