Dear John,
ExERSUG for some time has realized we need to initiate planning documents and to target our concerns to the Office of Scientific Computing (OSC), the Supercomputer Access Committee (SAC), and Program Offices, collectively, and to the management of the National Energy Research Supercomputer Center (NERSC). Washington has made it clear that they want ExERSUG to be an independent body, not being directed by OSC or others in DOE, thus having our stated positions available to DOE as a truly independent view from the users. We believe we can strengthen this process by contributing our views and the views of the users we represent on a frequent and timely basis.
The enclosed document, ``View from the Energy Research Users'' is ExERSUG's presentation of the users' collective views and concerns to OSC on issues we consider of paramount importance. These views are strongly held by a large fraction of our users. We make specific, carefully weighted, constructive policy recommendations, aiming at a more efficient use of existing financial resources. We urge you to give our document your serious attention.
The document makes a strong statement of support for keeping the Massively Parallel Processor (MPP) acquisition by NERSC on track. The NERSC MPP team has done a stellar job in structuring their procurement to get the latest technology that will do the job. But with delays being evident, especially in these budget cutting times, ExERSUG perceives that we need to make a strong statement in support of MPP procurement.
The document additionally argues for NERSC adding additional capacity, by way of cost-effective Symmetric Multi-Processors (SMP). The SMPs would act both as development platforms for the MPP, which have been relatively poor in software, and as cost-effective capacity engines that would allow offloading the C90, allowing better access by our established base of supercomputer vector codes, and additionally allowing expansion of the new-capability Special Parallel Processing (SPP) use of the C90. Furthermore SMPs with larger memories would allow new capability, true supercomputing large-memory computations not feasible on the C90.
Sincerely yours,
Jack A. Byers, Chairman, ExERSUG
Co-signees, members of ExERSUG:
Bastiaan J. Braams, Viktor K. Decyck, William B. Herrmannsfeldt, Brian E. Hingerty,
Stephen C. Jardin, Rick A. Kendall, Steven C. Pieper, Gerald L. Potter, Alan D. Turnbull
CC:
OER Program Directors:
Iran Thomas, (Acting) Associate Director, Office of Basic Energy Services, ER-10
Wilmot N. Hess, Associate Director, Office of High Energy and Nuclear Physics, ER-20
N. Anne Davies, Associate Director, Office of Fusion Energy, ER-50
Aristides A. Patrinos, (Acting) Associate Director, Office of Health and Environmental Research, ER-70
NERSC Managers:
C. William McCurdy, Director, NERSC
Michel G. McCoy, Deputy Director, NERSC
John M. Fitzgerald, Assistant Director, NERSC