Improvements to the indexing code could minimize that amount of lambdas to about 200 presented some energy, so it seems like one hundred million documents working with the whole 1,000 lambdas looks feasible, but additional possible it would be close to eighty million when performed.