By Calder Phillips-Grafflin
When the Wold building opens for classes next fall, it won’t just be a building with normal classrooms and labs; it will also be the home of the new Advanced Computing Lab (ACL), made possible by Union trustee John Kelley and collaboration with IBM.
Work on the Advanced Computing Lab dates back almost two years ago, when the planning for the Wold building was being finalized. At the time, John Kelley approached Union with the goal of providing high-performance computing resources to liberal arts colleges.
It wasn’t just going to be any old computing center: original plans for the room intended it to host an IBM Blue Gene supercomputer. Cooler heads soon prevailed, and for good reason: Blue Gene systems are very specialized and require complex programming and operation. It simply didn’t make sense to use a rare supercomputer as a tool for bringing computation to the liberal arts.
After months of discussion and planning, the High Performance Computing Lab is gone, replaced by the Advanced Computing Lab. Gone, too, is any idea of Union having its own supercomputer.
Instead, the ACL will combine high-end media workstations with a new cluster from IBM. The workstations will be standard 12-core dual-processor Mac Pro desktops running Linux, Mac and Windows, and will be used by students like any other computer lab. The cluster will be an IBM Intelligent Cluster, which combines Xeon server processors with NVIDIA computational graphics cards.
In theory, if a variety of issues can be solved, the new Advanced Computing Lab will be a wonderful resource for the college. However, that’s a big ‘if’: the issues include access control, maintenance, collaboration, planning, course development and institutional vision.
As it stands right now, very little has been done to meaningfully address these issues. Instead, the administration has, to paraphrase the words of the Computer Science department “put a very expensive cart before the horse,” focusing on the acquisition of computer hardware without planning for its long-term use and maintenance.
First and foremost is the problem of planning; there simply aren’t enough projects in the works to take advantage of the new cluster. Right now, the only people planning to use the full power of the cluster are the Neuroscience faculty, who will use its computational power to run simulations in manageable time.
Course development, too needs to adapt to the new resource. It would be a shame if Union didn’t offer CS and engineering courses that took advantage of the cluster and the ability to use computational graphics systems. These are both important areas of development in computing, and it would be idiotic not to teach them to Union students.
Last, but far from least, is the issue of who will maintain and control the new computing resources. Unlike the existing cluster, the new ACL will not belong to any department, and will instead be a shared campus resource.
Presumably assuming that few people will use the cluster, almost no planning has been done about managing access. Without a central authority controlling who can and can’t use the cluster, departments will likely use other resources on campus rather than spending their time trying to get time on the cluster.
The new Advanced Computing Lab will offer great promise to the campus community. If the administration can figure out constructive solutions to the numerous issues it faces, it will be a wonderful resource for departments and student research. If not, it will become little more than a white elephant to be shown off to wealthy donors and trustees. Let’s hope for the former.