$_ = '12d.9j7h.7tttt.125.ty.2.4.8.9f.y6.34gk.ll';

print "split: ", map "[$_]", grep $_, split /((?:\w+.){3}\w+)\.?/;
print "regex: ", map "[$_]", m< ( (?: \w+. ){3} \w+ ) \.? >xg;

__output___

split: [12d.9j7h.7tttt.125][ty.2.4.8][9f.y6.34gk.ll]
regex: [12d.9j7h.7tttt.125][ty.2.4.8][9f.y6.34gk.ll]
[download]

Using split we use the capturing feature and grep out the extraneous data to get the desired output. With the regex we do something similar by capturing all the groups in the string. So it would seem, in this case the regex approach would be better.

See. perlre and split for more info.
HTH

_________
broquaint

update: fixed code (wasn't dealing with the last group correctly)