. Author manuscript; available in PMC: 2013 Oct 1.

Published in final edited form as: Physiol Meas. 2012 Sep 26;33(10):1703–1715. doi: 10.1088/0967-3334/33/10/1703

Table 1.

Execution times for real valued (top) and complex valued (bottom) computation of the Jacobian matrix. Performance results are reported for an optimized routine on an Intel Xeon based PC, and for optimized routines using 1, 2 or 4 GPUs of an NVIDIA Tesla S1070 unit. Timing includes the transfer of data from CPU to GPU and transfer of the computed Jacobian matrix back to the CPU.

Computing Time - Real Valued Case
	CPU	1 GPU	2 GPUs	4 GPUs
Time [s]	297.5	14.2	8.3	5.6

Computing Time - Complex Valued Case
	CPU	1 GPU	2 GPUs	4 GPUs
Time [s]	745.4	35.4	20.8	14.7