First I'm going to take a chunk of the 1000 genome database. So I do,

 
plink --noweb --bfile /nas/data/GNlib/1000Genome/ALL/1000genome_all_merged --chr 6 --from-kb 20000 --to-kb 22000 --make-bed --out 1kg_chr6_20_22 --allow-no-sex

to get the chromosome 6 from 20 to 22 megabases.

From this small piece I want to calculate the LD matrix. Let me see, 2MB^2 = 4e12 bases. So I will end with a theoretical matrix of 4e12 elements. Must be a little less cause no all the positions are filled.