IBM CLuster

• The documentation is a little difficult to find but can be found here

It would appear to run jobs that request more than 2000 MByte memory, and then abort them after about 10 seconds.