Re: FPGA based database searching

> be used, slowing down the throughput. Target is a database searching of
> 262144 elements with 16 bit each in maximum 220 ms.

	If you are searching 16 bit datums then your search space is limited to  
65536 different datums. Why would you want to store more than this ? A  
hashtable comes to mind, for instance.

	What is your searching algorithm ? Why is it that complex ? Can you post 

a description ?

	Taking well-designed software as example, your 220 ms time budget is  
enough for Postgresql to perform about 3000-5000 simple SQL queries  
returning 1 row with btree index lookup on a table with millions of rows  
including network overhead (as long as it doesn't need to hit the disk),  
in 220 ms Xapian can make multiple full text phrase searches with ranking 

on a multi gigabyte text dataset. These are per core of course.

	So, I suggest reviewing your search algorithm ;)