Palm/FastFootprints
Palm/FastFootprints implements two method for rapid memory footprint access diagnostics. The two methods are whole-program and precise. Current footprint analyses can cause 200× or more slowdown with realistic inputs and are therefore impractical. The whole-program method reduces the overhead to 10% by computing upper bounds, but still yields inter-procedural insight through a call path profile. Our precise method uses additional static analysis and profiling to refine the upper bounds for intra-procedural loop nests.
Minos Computing Library
BigFlowSim
BigFlowSim is a workflow I/O simulator that captures key implementation choices for remote I/O, including intensity, reuse, locality, access pattern, and data movement. With BigFlowSim, it is possible to generate synthetic benchmarks, and then perform sensitivity analyses to quantify the effects of the various I/O parameters. In addition to generating synthetic workflows, BigFlowSim is also able to replay traces (of I/O access) captured from execution of actual distributed scientific workflows.
TListSpectrum
Time Series Data Utilities
OpenCGRA
Coarse-grained reconfigurable arrays (CGRAs), loosely defined as arrays of functional units (e.g., adder, subtractor, multiplier, divider, or larger multi-operation units, but smaller than a general-purpose core) interconnected through a Network-on-Chip, provide higher flexibility than domain-specific ASIC accelerators while offering increased hardware efficiency with respect to fine-grained reconfigurable devices, such as Field Programmable Gate Arrays (FPGAs). The fast evolving fields of machine learning and edge computing, which are seeing a continuous flow of novel algorithms and larger models, make CGRAs ideal target architectures to allow domain specialization without loosing too much generality. Designing and generating a CGRA, however, still requires to define the type and number of the specific functional units, implement their interconnect and the network topology, and perform its simulation and validation, given a variety of workloads of interest. This invention proposes OpenCGRA, the first open-source integrated framework able to support the full top-to-bottom design flow for specializing and implementing CGRAs: modeling at different abstraction levels (functional level, cycle level, register-transfer level) with compiler support, verification at different granularities (unit testing, integration testing, property-based testing), simulation, generation of synthesizable Verilog, and characterization (area, power, and timing).
PaKman: A Scalable Algorithm for Generating Genomic Contigs on Distributed Memory Machines
Conducting genome assembly at scale remains a challenge owing to the intense computational and memory requirements of the problem, coupled with inherent complexities in existing parallel tools associated with data movement, use of complex data structures, unstructured memory accesses and repeated I/O operations. To this end we have developed a tool, PaKman which presents a fully distributed method that tackles assembly of large genomes through the combination of a novel data-structure (PaK-Graph) and algorithmic strategies to simplify communication and I/O footprint during the assembly process. The algorithm deviates from the state-of-the-art de Bruijn graph-based methods and presents a novel perspective to addressing the assembly problem by incorporating: i) a novel distributed-memory data structure that enables contig enumeration with minimal coordination; ii) a novel contig generation algorithm with simplified I/O and communication patterns. Our results demonstrate the ability to achieve near-linear speedups on up to 16K cores (tested) on the NERSC Cori supercomputer; perform better than or comparable to other state-of-the-art distributed memory and shared memory tools in terms of performance while delivering comparable (if not better) quality; and reduce time to solution significantly.
Community Emissions Data System (CEDS)
The community emissions data system is a software system developed for research use, using the R open source programing language that produces consistent estimates of global air emission species over the industrial era (1750- latest full year) by country, sector, and fuel. The system focuses on aerosol and aerosol precursor emissions, e.g. emissions that impact atmospheric chemistry and aerosol loadings. Emissions are also mapped to spatial grids, produced in netCDF format, for use by Earth system and other atmospheric models. Data inputs to this system include emission inventories and driver data, such as energy consumption and population estimates, and default emission factors, which are used together to produce consistent estimates of emissions over time. The data system is unique in its blending of emission data from multiple sources together with driver data, to consistently estimate emission trends over time in an open source framework. This software was planned, from original conception and in the project proposal and FWP, to be released as open-source software.