List, A memory question .. on Linux 2.6 i connect with mapfraw to a pdl byte file with 8000 columns and 300000 rows in order to collect column statistics, slicing one row at a time and then undef'ing it. As the loop runs, physical memory grows until it equals the virtual memory (as seen with 'top'). To limit physical memory I tried to reconnect for every 10000 rows, but that causes a segfault, however that works for smaller files (maybe because 2 large mapfraw's exceed max memory due to subroutine scope). Have I overlooked a way to plow through large pdl files while freeing memory for data that will no longer be visited? can files larger than available ram be sliced/accessed quickly?
You can tell that I am no PDL-buff .. i just use a small subset of its features and have put them in a library that has worked good for a long time, but now data sizes are growing quickly and I wonder about size limits .. Greetings, Niels L _______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
