android - Compact alternatives to Java ArrayList<String> -
this question has answer here:
i need store large dictionary of natural language words -- 120,000, depending on language. these need kept in memory profiling has shown algorithm utilises array time bottleneck in system. (it's spellchecking/autocorrect algorithm, though details don't matter.) on android devices 16mb memory, memory overhead associated java strings causing run out of space. note each string has 38 byte overhead associated it, gives 5mb overhead.
at first sight, 1 option substitute char[] string. (or byte[], utf-8 more compact in case.) again, memory overhead issue: each java array has 32 byte overhead.
one alternative arraylist<string>, etc. create class same interface internally concatenates strings 1 gigantic string, e.g. represented single byte[], , store offsets huge string. each offset take 4 bytes, giving more space-efficient solution.
my questions a) there other solutions problem low overheads* , b) solution available off-the-shelf? searching through guava, trove , pcj collection libraries yields nothing.
*i know 1 can overhead down below 4 bytes, there diminishing returns.
nb. support compressed strings being dropped in hotspot jvm? suggests jvm option -xx:+usecompressedstrings isn't going here.
i had develop word dictionary class project. ended using trie data structure. not sure size difference between arrraylist , trie, performance lot better.
here resources helpful.
https://en.wikipedia.org/wiki/trie
https://www.topcoder.com/community/data-science/data-science-tutorials/using-tries/
Comments
Post a Comment