Quite honestly ... you're just going to have to re-think your strategy here. Your "100GB of sequences" is very-obviously a database, not any sort of reasonable "in-memory data structure," no matter how much memory your machine has, nor how many simultaneous threads (why?) you are now attempting to run.