The hpg1-compute partition is being retired and should no longer be used when submitting jobs. The SLURM scheduler will now reject jobs that list the hpg1-compute partition and it should be removed from job scripts.
In general, users should submit jobs without specifying a partition unless there is a viable reason to do so. The compute nodes that were a part of the hpg1-compute partition are now in the hpg-default partition.
Please open a support request if you have problems following this change.