As you might have experienced, the usage of the hugemem nodes (1TiB RAM) was expensive, due to a high Processor Equivalent (PE) count for every single run. Now the factor to compute the PEs for the hugemem node has been reduced to allow the usage of the resources at the same price of the normal nodes.
This will results in much cheaper run on the hugemem nodes!
More info here (search for "PE" in the page)
Happy Computing!
Mangement@TSD