binary into matlab workspace

I have a large number of bin files that I'm converting and importing the data into the matlab workspace into variable X for processing. Data is a column of numbers.

X=convert(i,pwd);

convert is a function which converts binary files in pwd directory and pulls out column i from each file and concatenates and puts into X.

What I'd like to do is

when length(X) = length(files(pwd)/2; or some value set by me

Create X1=convert(i,pwd); <-- but start i where X left off and not from begining of bin file. And so on for X2

So Id like to basically split X into X1 X2 .... which are individually small enough to fit into memory and then process them.

If I could somehow pause the execution and redirect where convert stores data without disturbing anything else.

Answer 1

[Walter Roberson]

Sorry, you are going to have to "disturb" your code interface to handle that. For example, you could have convert() output a second output that was the ftell() value of the place it left off reading, and convert() could take an second input that would be a maximum length, and a third input that would be an fseek() position to start from.

Putting the length constraint to the actual reading should not be difficult; there is an optional size parameter for fread().

[Baba]

This is my getchan function:

function [DATA] = convert(chan,wd)

% Search for .bin files in the wd directory

DATA=[];

% Loop over every .bin file and build up DATA

for i=1:length(files)

% Make sure we're reading from the wd directory

[d] = readbin (fullfile(wd,files(i).name));

% Append the first column of what we read to DATA

data=(d(:,chan));

DATA=[DATA; data];

end

end

[Walter Roberson]

Yes, you are definitely going to need to "disturb" that interface. You want to break the data when "number of files divided by 2" *samples* have been read, even though each file might contain thousands of samples. For example if there were 10 files in the directory, you have asked to break after 10/2 = 5 *samples*, which would be like 2048 variables if each of the 10 files contained 1024 samples.