as I already said in another post on your thread, performance is probably completely irrelevant for the small dataset your are talking about.

However, just in case you are interested, I ran a detailed benchmark on a very similar problem a bit less that a year and a half ago. The results are here: Re: Performance problems on splitting long strings. You'll see that unpack won the race, but substr wasn't that far behind.

It did make some difference to me, however, because I was running the processing of two 6-GB files, with the long string to be split representing at least 75% to 80% of the data volume.

This was just for your information. Again, I don't think you should care at all about that for your low data volumes.