On Mac OS 10.8 I tried the Hilbert transform on a source waveform matrix of 2503 dipoles x 52000 time samples. It still indicates 0 percent complete after 15 hours. I have 4 GB of RAM. I can run the Connectivity processes on this data with no problem, and it does not seem that the Hilbert process would need any more RAM then the Connectivity processes.
I am trying the Hilbert on a Linux system, but keep getting aborted for perhaps several reasons. I requested 16 GB on the last qlogin attempt, but that did not finish before I went past the requested time limit. How long should this take to calculate?
-Jeff