Compound clustering/distance matrix calculation

I am writing a protocol to cluster a library of ~70k compounds by calculating Indigo fingerprints then trying to calculate the distance matrix.  That's where things are not working.  Is this too large a data set to try to cluster?  Is there a more efficient method to determine chemical diversity using the KNIME nodes? 

70k is a lot. I find 5k is around the limit.
You could also cluster by other means such as by Murcko scaffolds, or you could also use the universe marker and neighbour gram approach with your fingerprints.
Another option is to split out the fingerprints into binary columns with the erlwood BitVector node and then use more traditional clustering tools such as the k-means and k-medoids nodes.
Also you have the moss fragment node to cluster molecules based on fragments.

Simon.