For our planning purposes, what computer resources will likely be available to PSAAP Alliance Centers during the expected contract period, and how many processors must our simulations scale to effectively use?
The answer can be framed in terms of experience with the current ASAP Centers as well as Tri-lab plans. It must be recognized that there is uncertainty due to available funding and other factors as to exactly what architectures and computing power will be available and when. The Alliance Centers will only have access to systems in the Open partitions. We have and will continue to provide a meaningful fraction of the total ASC-funded computing resources (Closed and Open) in the Open on a regular basis (order 5%) as well as opportunities for scheduled dedicated time on the available Open systems. With the current ASAP program, we have examples of ASAP runs using over 10M processor hours on the IBM BG/L system when it was in the Open partition at LLNL. Some systems such as the SNL Red Storm system may have an architecture such that they may be able to swing the whole system from Open to Closed and back; others such as the Purple system at LLNL or Q at LANL may not have such an architectur
