Re: [gui-dev] QueryRouteTable.java.patch (second version)

Philippe Verdy wrote:

> From: "Zlatin Balevsky" <[email protected]>
>
>> A small note on RLE - we could relatively easily modify our table 
>> reading code to read directly from a RLE'd table; this alone could 
>> save us a lot of memory on the leaf tables.  This seems like the 
>> easiest first step to take.
>
> <...>

> The more we increase the virtual bitsize, the more the RLE compression 
> will become effective to maintain (or not increase significantly the 
> average compressed size).


Lets hold this thought for a moment - suppose we want to use RLE only 
for the tables of leaves.  We could make their virtual size some large 
number like 1MB - if it is a single bitset it will compress with RLE 
wonderfully.  And since leaf tables tend to change rarely so we are not 
going to re-allocate storage too often.  The reading complexity is 
proportional to the fill rate and can be implemented as a simple 
iterative read of the compressed table.

While this may not be a great improvement, it will reduce the number of 
false positives sent to leaves greatly with minimal increase in cpu 
usage and virtually no increase in memory usage.  Its certainly worth 
experimenting with ;)

_______________________________________________
gui-dev mailing list
[email protected]
http://www.limewire.org/mailman/listinfo/gui-dev