Huge Data file Import Export

I have a huge data file as .txt. Each element is separated by a ; and each row by a new line character. The data has been generated by a c++ program. Now, the no. of columns in each row is not constant. I need to read each row in MATLAB. I converted the file to .xlsx and used the following code:

    A=xlsread('file.xlsx','Sheet1','2:2');

The command takes a lot of time to execute. The data is 30,000 X 500 approx.

1. How do I read the data in a faster way? I tried using csvread but could not implement it.

2. How do I implement the xlsread code in a loop? I need it to run from '2:2' to '30000:30000'. If I import the whole data, it causes memory failure.

3. In what format should I generate data from c++ so that it is fast to import in excel?

[Cedric]

A 3E4x3E4 matrix size in memory is 7.2GB if stored as double.

These huge files should be stored/managed using a binary format (your own, HDF5, netCDF, etc) if you wanted to be efficient.