The rapid progress of computing technology and rapid
turnover creates negative environmental impact through
computing equipment e-waste. Because the growing computing
capability is of critical importance to science, it is
critical to find solutions. This NSF project's objective is
to create extended lifecycle pathways for
scientic computing equipment, and thereby increase
scientific computing productivity per unit e-waste and
carbon emissions.
The opportunity to create extended life with reduced carbon
emissions depends on the Zero-carbon Cloud Project's novel
approach leveraging the growing excess renewable power power
grids for low-cost power and low-cost operation.
We are creating economic models of the extended lifecycles and broader
ecosystem, documenting the significant economic incentives
and potential positive impacts on e-waste and carbon
footprint. We will connect scientific computing facility
operators (universities, national laboratories, and more)
with innovative new providers. Together, we will solve the
practical challenges, and publish best practices and
standard models for extended lifecycle engagements. This work is supported by the US National Science Foundation under NSF Grant OAC-2019506.
Publications and Reports
- (NEW!) Andrew A Chien and Mark Dietrich, Navigating Dennard, Carbon and Moore: Scenarios for the Future of NSF Advanced Computational Infrastructure. , ACM Conference on Practice and Experience in Advanced Research Computing, July 2022, Boston, Massachusetts. slides . Also available as UChicago CS-TR-2022-03 (April 2022).
- Mark Dietrich and Andrew A Chien Dennard, Carbon, and Moore: Technology Trends Shaping Future Investment in Scientific Computing Equipment , UChicago CS-TR-2022-04, (May 2022)
- Mark Dietrich and Andrew A Chien Options for Extending the Life of Scientific Computing Equipment , UChicago CS-TR-2020-08, (October 2020)
- ACM PEARC20 BoF on "Extending Lifetime" Presentations and Vendor information (July 2020)
- Mark Dietrich and Andrew A Chien, Analysis of US National Science Foundation (NSF)
Awards for Scientific Computing Equipment (SCE),
2013-2019. Available as a UChicago Technical
Report, CS-TR-2020-04 , July 8, 2020.
- Birali Runesha, Mark Dietrich, and Andrew A Chien, Scientific Computing Equipment (SCE) Decommissioning
in the University Community, 2017-2023: A Survey of the
Coalition for Advanced Scientific Computing (CASC), Available as a UChicago Technical Report CS-TR-2020-05, July 8, 2020. and
presentation slides.
Total Cost of Ownership and Economic Modeling
Computing is critical for science (discovery), education and commerce, but there is increasing concern about the costs, climate impact and environmental impact of those capabilities. Today, many organizations follow a model, or usage pathway, of buy new, operate for five years and then decommission. We hope to identify alternative models that deliver greater scientific capability at lower cost to inform future practice and government investment.
Specific questions being studied include:
- What pathways and lifetimes deliver the lowest-cost per compute capability?
- How do these approaches affect environmental impact? (scope 1-3)
- How do new options to operate equipment at low-cost, low-carbon data centers create new opportunities?
This is supported by the US National Science Foundation under NSF Grant OAC-2019506 as part of the Extending the Lifetime of Scientific Computing Project.
Data Collection: The Zero Carbon Cloud team is reaching out to a number of government funded computing facilities seeking the following information for computing systems:
- Capital costs
- Technical specifications
- Physical size and weight
- Power costs
- Space costs
- Information about power sourcing.
This data will be aggregated and used to perform analysis of current and potential pathways and lifetimes. No individual responses will be released, and care will be taken to avoid compromising individual responses and potentially sensitive information in aggregated results analysis.
For more information, contact Prof. Andrew A Chien and Mark Dietrich .
Events
- BoF at PEARC20
(7/29/2020): How to
Extend the Productive Lifetime of Scientific Computing
Equipment (Chien, Runesha, Dietrich, and Gardner)
(community outreach for the opportunity)
- Extended Lifetime Vision (Chien)
- Decommissioning SCE Survey Results (Runesha)
- Operating Intermittent Resources at High Productivity (Gardner)
- Datacenter and Vendor Opportunities (Dietrich)
- Teams Workshops on Extended Lifetime for SCE
(7/31/2020, 2pm CT). (Chien and Dietrich) (refine working
models)
- Registration for the Teams Workshop (closed).
This workshop will have small working groups to identify challenges and work on solutions to
make each of the models viable for a broad swath of the academic scientific computing community.
- Kickoff and Charge
- Working Groups - Reliable Green Colo, Intermittent Green Colo 1, Intermittent Green Colo 3...
- WG Output Models: Reliable, Greeen Compute (RGC); Intermittent, Green Compute - hardware (IGC1); Intermittent, Green Compute - services (IGC2)
- Final Report - Analyzes costs for all three models is now available! (10/22/2020)
Publications and Reports
- Analysis of US National Science Foundation (NSF)
Awards for Scientific Computing Equipment (SCE),
2013-2019. (Dietrich and Chien), Available as a UChicago Technical
Report , July 8, 2020.
- Scientific Computing Equipment (SCE) Decommissioning
in the University Community, 2017-2023: A Survey of the
Coalition for Advanced Scientific Computing (CASC)
(Runesha, Dietrich, and Chien), Available as a UChicago Technical Report , July 8, 2020. and
presentation slides.
More information on the Zero-carbon Cloud project can be found at:
2019 Overview and
2015 vision.
Questionaire: Assessing the extended-lifetime Opportunity for Scientific Computing Equipment (April 2020)
Please click HERE (link deactivated, June 2020) to fill out the Scientific
Computing Equipment Opportunity questionaire.
Your responses will be held in confidence and aggregated to create a projection of the potential scale, economics, and scientific benefits of extended lifetimes.
The full survey report is available , July 8, 2020.
The Large-scale Systems Group
(LSSG)
is at the
University of Chicago's
Department of
Computer Science
LSSG is also affiliated with Argonne National Laboratory's
Mathematics and Computer Science Division.
|