Keys are simply short strings of integers.
Hm. Does that mean each key is a single short integer, or each key contains several integers?
Here's why I ask. The following build two hashes with 1 million key/value pairs.
- In this one the keys are strings containing 3 x 7 digit integers; and the values are 10 x 7 digits integers delimited by spaces:
C:\test>p1
Perl> #5.6MB ;;
Perl> $h{ "$_ " x 3} = "$_ " x 10 for 1e6 .. 2e6;;
Perl> # 299MB ;;
1 million key/value pairs 300MB
- Same thing, but packing the integers to 4-byte binary values:
C:\test>p1
Perl> # 5.6MB ;;
Perl> $h{ pack 'V3', ($_) x 3} = pack 'V10', ($_) x 10 for 1e6 .. 2e6;
+;
Perl> # 237MB ;;
Voila! You've stored the same information and saved 1/5th of the space.
If the numbers can be stored as shorts or bytes, you can save even more.
I'm also having trouble reconciling your OP mention of a 3GB hash with the description of key/value sizes and total numbers above?
This creates a 3 million key/value pair hash with 21-byte keys and 80-byte values and the total memory requirement is <1GB:
Perl> $h{ "$_ " x 3} = "$_ " x 10 for 1e6 .. 4e6;;
Perl> # 912MB ;;
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
|