You are trying to re-invent a database and query engine.

Keeping all this data in memory will not work unless you have a computer with a huge memory to avoid the repeatedly swapping in and out of your data and even then you still have to write a program to efficiently search through all those arrays.

I would do it as follows:

1. Read each file into its own database table and index the field with the timestamp. You cannot make the timestamp the key of each record as you seem to have multiple records with the same timestamp in each file. Use a module such as Date::Parse or DateTime::Format::DateParse to turn the timestamp into a standard format that will be the same in all reocrds.
2. Use standard SQL to match the timestamps in your FILEA-table with the timestamps in the FILEB-, FILEC-, ... tables.

Alternatively but only if all the files have timestamps in strict time-sequential order, you can open a filehandle to each of the files and on a line-by-line basis, loop through FILEA, extract the timestamp, transform the timestamp into a standard format and then iterate through all other files checking their timestamps (after transforming those also into the same standard format) and output the record when the timestamps match until you hit a timestamp past the timestamp of the main loop. Then you do the same for the next FILE until all FILEs have passed the timestamp of the main file and you go to the next record in the main file.