Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Pavan Balaji Project Name: radix Division: MCS Project title: Scalable Parallel System Software Associated funding: DoE core, SciDAC, NSF, FASTOS Other Systems: breadboard (unlimited) ALCF Mira/Cetus/Tukey (50,000,000 cpu hours) Science: Our mission is to develop the technologies required to dramatically increase the productivity of scientists developing applications for parallel supercomputers. The focus is fourfold: integration of parallel programming tools, reuse of parallel program components, development of scientific computing toolkits and portable libraries, and exploration of requirements of future parallel computers. Project description: This allocation will be used for optimizations to various systems software components: (1) programming models and runtime systems, (2) data I/O and file systems, (3) fault tolerance, (4) operating systems, and (5) data analysis and visualization. Primary goal of the project is to perform research on improving the core systems software on high-end computing platforms. The radix allocation covers a large number of cross-cutting projects involving subsets of researchers in the group. Three core projects are presented here, but some of the techniques used are shared between them and are not easily separable (e.g., MPI improvements or resilience improvements): 1. Fault Tolerance Techniques: - Local memory management * SDC Detection through Data Monitoring + Detect outliers on application data through one-step ahead prediction * Energy Efficient Multilevel Checkpointing + Power monitoring integration into resilience libraries and energy-aware ckpt. scheduling * I/O Prediction for Multilevel Checkpoint + Inter-application coordination and I/O collision avoidance - Distributed memory management * MPI ULFM + Improvements to MPICH to deal with faults * Resilient-MPI and FTI Integration + Experimentation and evaluation of FTI coupled with resilient-MPI 2. Low-level Hybrid Programming with MPI and threads: - Communicating with Massive Threads * Integrate threads with MPI to support communicating with light-weight threads - Recently funded project (Argo) that deals with improvements to both MPI and the threading model * MPICH improvements to play well with user-level threads * OS improvements to provide light-weight threads that "understand" MPI concepts 3. Data I/O, RMA and Active Messages: - Large-Scale Storage System Prototype and Simulation * Evaluate future storage system using discrete event simulation with MPI and InfiniBand - Dynamic Locality-Aware Workflows * Dynamically distribute the tasks among nodes using Asynchronous Dynamic Load Balancer - Parallel I/O autotuning * Automate the selection of good configuration parameters for HPF5, MPI-IO and the parallel file system Industry partnership: Project URL: http://www.mcs.anl.gov/group/pmrs/ Current FY Hours Used: undetermined amount New FY Requested allocation: 2000000 Q1: 500000 Q2: 500000 Q3: 500000 Q4: 500000 Justification: 2M core hours requested: Three core projects. All projects have been demonstrated at scale in the previous years and form the core of the system software stack used in production on the Fusion and Blues machines. We expect to run around 75 studies for each project on Blues and Fusion, scaling from 1 node to 256 nodes (average runtime of 30 minutes) - Blues: 3 (projects) x 75 (studies) x 512 (1 to 256 nodes) x ½ (hour) x 16 (cores) = 920K - Fusion: 3 (projects) x 75 (studies) x 512 (1 to 256 nodes) x ½ (hour) x 8 (cores) = 460K Some smaller projects: We expect to run around 75 studies for all projects, scaling from 1 node to 256 nodes (average runtime of 30 minutes) - Blues: 75 (studies) x 512 (1 to 256 nodes) x ½ (hour) x 16 (cores) = 310K - Fusion: 75 (studies) x 512 (1 to 256 nodes) x ½ (hour) x 8 (cores) = 155K A small fraction of this would be used for debugging. Storage requirements: No large storage requirements, since the data generated will not be persistent. Thank You, The LCRC Accounts System