[continue] Make batch File wont work on 128 and 256 images

[teknosains]

Haloo, from the issue i posted here https://github.com/dnouri/cuda-convnet/issues/2 , there is a typo the images shud be 14thousands not million , sorry.

the error message when i ran the make-batch code for 128-256 was not showing, it just 'freeze' my computer ( hang ). and hit almost 4GB of ram

fyi my computer specs : 4GB of ram , ubuntu 12.04

i tried to trace line per line of this ( noccn/dataset.py) , and it was 'freeze' when the code executed these line : # i add this : print "FInish 2" to trace in which lines the code terminated print "FInish 2" # <----- printing dot (.....) this this row is printed,,and killed data = np.vstack([r for r in rows if r is not None]) #<---- I thought this line could not handle large numpy returned in 'rows' print "finish 3"

---

I tried to change the PIL Image Open method with 'imread' which is stated faster than PIL but still got the same.

do you have any suggestion ?

thanks