Phase 0 of the cluster contains “big memory” nodes, each with a large amount of RAM. See the output of whatsfree for the configuration (amount of cores/RAM available in each of these nodes).

To use these nodes, you must submit jobs to the special bigmem queue. For example, in your job script you can include the #PBS -q bigmem directive:

#PBS -N bigexample
#PBS -l select=1:ncpus=24:mem=494gb
#PBS -q bigmem

For interactive jobs, just specify -q bigmem along with your qsub command:

$ qsub -I -l select=1:ncpus=24:mem=494gb -q bigmem