Innovation in computer hardware, ranging from the simplest consumer devices to the most powerful scientific computers, relies on availability of software and hardware design and evaluation tools. The development of the cyberinfrastructure is itself a significant undertaking due to the complexity of today's computers and applications. This complexity has lead to ad-hoc infrastructure that is difficult to quickly extend in new ways, minimally tested, and arduous to use for comparing innovations. This burden greatly impedes progress of computer hardware research and development. The community-driven creation and maintenance of robust cyberinfrastructure can overcome this challenge by leveraging and combining effort and investment made by multiple groups. This workshop sets a foundation to grow a community for developing more reusable, better tested and interoperable tools. The energy, performance, reliability and cost of information technology will be further improved at a faster pace as a result.

The cyberinfrastructure for computer architecture (hardware) includes software simulation, hardware emulation, analytic models and benchmarks for CPU/cache, network-on-a-chip (NoC), memory and storage. These tools are used to analyze existing systems and their bottlenecks, develop new hardware capabilities, and study and evaluate design trade-offs. While the tools are deeply ingrained and fundamental to computer architecture, the architecture cyberinfrastructure suffers badly from fragmented, ad hoc and disparate efforts, arising from research expediency and development burden. To address the situation, this workshop brings together tool developers and users to assess the current state of the cyberinfrastructure and to establish strategies and build a community for more coordinated and joint development. The workshop develops 1) an assessment of the current state of the cyberinfrastructure; 2) roadmaps of needed tool capabilities and support for future technologies; 3) ways to leverage and combine disparate development effort by multiple groups; and 4) a methodology to establish and sustain a cohesive community that builds and supports a set of open-source, extensible, scalable, validated and interoperable tools. The workshop outcomes serve building a community framework for simulating computer architecture.

Project Report

A framework of interoperable simulators for computer architecture is long overdue. It is important to bring together and build a community of researchers and implementers of architecture simulation, emulation, and modeling tools for memory, CPU/cache, storage and the network on a chip. This project engaged academia, government, and industry to work openly and synergestically as a community on simulation tools and experiments. It created a set of recommendations for the community to create interoperable and validated simulation infrastructure and reproducible experiments. Through a series of community events, including a workshop of invited participants, this project 1) assessed the current state of architecture simulation and hardware emulation infrastructure; 2) considered the priorities for future work on interoperable and validated simulation frameworks; 3) identified ways to leverage investment by multiple groups in simulation frameworks and the experiments conducted with them; and 4) began building a community to support the infrastructure. From the project activities and events, the effort made several recommendations for building interoperable and validated simulation frameworks and reproducible experiments: 1) a means for governance of consensus and decision making needs to be established; 2) a central repository should be developed to hold simulation artifacts (simulators, emulators, data sets, benchmarks) and experimental results; 3) the community should leverage investment among all groups, and 4) the simulation infrastructure should build on an existing foundation and incorporate current simulators.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Communication Foundations (CCF)
Type
Standard Grant (Standard)
Application #
1148646
Program Officer
Almadena Chtchelkanova
Project Start
Project End
Budget Start
2011-08-15
Budget End
2013-07-31
Support Year
Fiscal Year
2011
Total Cost
$89,335
Indirect Cost
Name
University of Pittsburgh
Department
Type
DUNS #
City
Pittsburgh
State
PA
Country
United States
Zip Code
15260