Table II.

Comparison in CPU time between L-LF and F-LL schemes. Ionic sub-steps are uniform, except for the last row, where Δtau; is adapted asynchronously for each node independently following [13].

Method	Δt (ms)	Δt (ms)	Conduction Velocity (m/s)	CPU time (s)
	0.1	0.1	0.522	16.69
L-LF	0.1	0.05	0.532	27.72
	0.1	0.01	0.555	123.76
	0.1	0.1	0.536	17.06
F-LL	0.1	0.05	0.549	27.64
	0.1	0.01	0.561	131.75
	0.1	0.01–0.1*	0.543	18.113

^*Last row, uses adaptive time-stepping, which is possible for F-LL but not for L-LF (see main text).