I added W-BER to my benchmark. It just about beats binary ('v*') on space. About the same for unpacking but slower for packing.
Any way you slice it, the single best option is to restructure the schema to cause less data to need retrieval...but then you know that :)
## Note this run uses 1000 documents rather than 5000. ## The relative performance remains the same though. Run with 15000 unique words in 1000 documents (Ave: 554 words/doc) ASCII uses 4755001 bytes W-BER uses 3196726 bytes Binary uses 3279128 bytes Elias uses 4118678 bytes 1 trial of Packing ascii (9.078s total) 1 trial of Unpacking ascii (2.942s total) 1 trial of Packing W-BER (16.144s total) 1 trial of Unpacking W-BER (8.219s total) 1 trial of Packing binary (9.504s total) 1 trial of Unpacking binary (8.223s total) 1 trial of Packing Elias (11.176s total) 1 trial of Unpacking Elias (9.750s total)
In reply to Re^6: Byte allign compression in Perl..
by BrowserUk
in thread Byte allign compression in Perl..
by MimisIVI
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |