fast alternative to continuous wavelet transformation cwt

Howdy. I am using cwt to analyze time series data and this portion of the code, revealed by the profiler, is making my analysis computational prohibitive. Currently, I'm looking at about 250 scales and the data vector is of length 1e6. Ideally, I'd like to increase that an order of magnitude (to about 4e7) and still do the analysis in a reasonable amount of time. Any ideas on how to speed things up?

Answer 1

[Wayne King]

Are you using cwt, or cwtft to do the CWT? cwtft uses the Fourier transform and might be faster. Again you would have to profile it.

Answer 2

[Dan]

Thanks Wayne, I am using cwt. I've seen the cwtft but am still using R2010a, could make the switch if cwtft is truly faster. Otherwise, I was curious if there was something simple overlooked, like vector length for the old radix-2 FFT..