CME Science Process

• Science proposals • Allocation proposals for cycles • Allocation agreements for storage (informal?) • Allocation agreements for software

engineering (formal proposals, e.g. XSEDE, as well as informal negotiations?)

SCEC HPC Special Projects1. SI2

– NSF (OCI, Geo) funded - 3 year (Sept 2012- Aug 2015)2. Geoinformatics

– NSF (Geo) funded - 2 year project (Oct 2012-Sept 2014)3. Geoinformatics – Phase II

– NSF (Geo, OCI) funded - 2 years (May 2014 – April 2016)4. Broadband Platform

– PGE, SCE funded - 1 year (Aug 2013 – July 2014)

SCEC Annual Allocations1. USC HPCC

– Linux Cluster (some large memory nodes) and 6 GPU Nodes– Current Award: 1.5M SU with 30TB disk storage

2. NSF XSEDE– Linux Clusters, Shared Memory Computers, GPU nodes– Current Award: 55M SU with 50TB disk and 200 TB tape– O FTE Support

3. DOE INCITE– Leadership-class Cray Supercomputers some with GPU nodes– Current Award: 101M SU with 297TB disk and 148 TB tape– 1 FTE Support

4. Blue Waters– NSF Track 1 Cray cluster with GPU nodes accessible in Jan 2013– Current Award: 3M Node hours with 50TB disk storage– 0 FTE Support

CME Science Process

• examples to support the general point that this is all very complicated.

NSF Task Force on Software for Science and Engineering

The Task Force on Software for Science and Engineering has formed the following major recommendations on how NSF can best support the research, development, and maintenance of software infrastructure:

1. NSF should develop a multi-level (individual, team, institute), long-term program of support of scientific software elements ranging from complex applications to tools of utility in multiple domains. Such programs should also support extreme scale data and simulation and the needs of NSF’s Major Research Equipment and Facilities (MREFC) projects.2. NSF should take leadership in promoting verification, validation, sustainability, and reproducibility through software developed with federal support.3. NSF should develop a consistent policy on open source software that promotes scientific discovery and encourages innovation.4. NSF support for software should entail collaborations among all of its divisions, related federal agencies, and private industry.5. NSF should utilize its Advisory Committees to obtain community input on software priorities.