BrowserUk's unpack approach is probably a bit faster, but here's the 'standard' regex approach:

>perl -wMstrict -le
"my $s = 'AACCCDGYAEELPSWWYA';
 ;;
 for my $n (2 .. 5) {
   my @subseqs = $s =~ m{ (?= (.{$n})) }xmsg;
   print qq{n $n: @subseqs};
   }
"
n 2: AA AC CC CC CD DG GY YA AE EE EL LP PS SW WW WY YA
n 3: AAC ACC CCC CCD CDG DGY GYA YAE AEE EEL ELP LPS PSW SWW WWY WYA
n 4: AACC ACCC CCCD CCDG CDGY DGYA GYAE YAEE AEEL EELP ELPS LPSW PSWW 
+SWWY WWYA
n 5: AACCC ACCCD CCCDG CCDGY CDGYA DGYAE GYAEE YAEEL AEELP EELPS ELPSW
+ LPSWW PSWWY SWWYA
[download]