This will generate 10 mat files, 9 of which contain a 1024x1 variable of contiguous samples from your original mat file and 1 of which contains the remaining 704 samples:
fn = '64qam12';
S = load([fn '.mat']);
block = 1024;
N = ceil(numel(S.var5)/block);
for i = 1:N
var5 = S.var5((i-1)*block+1:min(i*block,end));
save(sprintf('%s_%d.mat',fn,i),'var5');
end
Alternatively, this wil save 10 mat files, each of which contains a 1024x1 matrix, the last of which is padded with NaNs on the end:
fn = '64qam12';
S = load([fn '.mat']);
block = 1024;
N = ceil(numel(S.var5)/block);
S.var5 = [S.var5; NaN(N*block-numel(S.var5),1)];
for i = 1:N
var5 = S.var5((i-1)*block+1:i*block);
save(sprintf('%s_%d.mat',fn,i),'var5');
end
