On a Windows XP PC with 1 gigabyte of main memory and a Pentium IV, I can sort two million 24-bytes strings in just over half a minute, but trying to sort four million caused major thrashing (the whole machine slows to a crawl) followed by Perl dying with "out of memory."

Fifty thousand filenames shouldn't be a problem on reasonably recent machines.

perl -wle '@a=(1..2_000_000); @b=sort map {scalar gmtime $_} @a; print
+ "@b[0,1,2,1_999_997,1_999_998,1_999_999]"'
[download]

Fri Jan  2 00:00:00 1970 Fri Jan  2 00:00:01 1970 Fri Jan  2 00:00:02 
+1970 Wed Jan 21 23:59:57 1970 Wed Jan 21 23:59:58 1970 Wed Jan 21 23:
+59:59 1970
34.42user 0.32system 0:35.76elapsed 97%CPU (0avgtext+0avgdata 29949952
+maxresident)k
0inputs+0outputs (122789major+0minor)pagefaults 0swaps
[download]

if there are say 50,000 filenames then will that array takes those names??

Perl arrays are implemented as arrays of C pointers so you'll probably run out of memory for the data in it before you reach the maximum length. Theoretically a perl array has a 2**31 = 2147483648 element limit. On a 32 bit perl / os you'll run out of memory for the array itself somewhere before half that amount.

update: look up "AV" in http://gisle.aas.no/perl/illguts/ if you're really interested.

The correct answer to "what is the maximum size of the array?" is "lots:" it will depend on the size of the elements (an array of bytes can have more elements than an array of 1 MByte MPEGs) and how much memory is available at that time.

To answer your specific question: it will probably easily handle those 50,000 names, but this is also tending towards the neighborhood where using a database will be a Good Idea.