prior.tex

\noindent \textbf{Prior NSF Support.} PI Cong Liu  has recently received the following NSF grant:

\noindent{\bf CSR: Small: Predictable Real-Time Computing in GPU-enabled Systems} (CNS-1527727, PI: Cong Liu, 10/15-9/18, \$354,076)

This proposal addresses the challenge of applying relatively unpredictable GPU hardware resources in
safety-critical systems, which ensures predictable real-time correctness.

\noindent {\em Intellectual Merit:} A set of new GPU-aware
resource allocation algorithms and the associated timing validation tools that yield quantifiable guarantees on
real-time correctness will be developed. Moreover, an ecosystem of GPU
resource management that enables GPUs to be predictably utilized in a
real-time multi-tasking environment will be built at the OS level. 

\noindent {\em Broader Impact:} This project has just begun.  The grant will be used to support two PhD students (including a female student) for three years.

\noindent{\bf XPS: FULL: CCA: Collaborative Research: CASH: Cost-aware Adaptation of Software and Hardware} (CCF 1439156, PI: Hoffmann, 09/2014-08/2018, \$300,000)\\
%\noindent{\bf EAGER: HAWKEYE}: A Cross-Layer Resilient Architecture to Tradeoff Program Accuracy and Resilience Overheads <HANK: Fill in details>\\
\noindent{\bf CNS: SMALL: BreezeFS} \\
\noindent{\bf II-New: RIVER: A Research Infrastructure to Explore
  Volatility, Energy-Efficiency, and Resilience} (CNS 1405959, PI:
Chien, Co-PIs: Foster, Gunawi, Hoffmann, Scott, 07/2014-07/2017,
\$997,432)

All projects advance the study of adaptive computing systems.  CASH
studies the modeling and allocation of fine-grain hardware structures
to produce cost savings in infrastructure as a service (IaaS) clouds.
BreezeFS studies adaptive management of multi-store systems for
clouds.  Among other things, RIVER's infrastructure supports study of
large-scale adaptation to meet power and performance goals in
distributed computing systems.  

\noindent {\em Intellectual Merit:} In the CASH project we have
developed novel hardware structures for performance monitoring as well
as novel combinations of machine learning and control theory for
allocating fine-grain hardware resources and minimizing costs.
BreezeFS has just begun, but it will study application of adaptive
tradeoff management to storage systems.

% --------------
\noindent {\em Broader Impact:} The CASH project has already resulted
in four pulbished papers from the University of Chicago
\cite{kim-cpsna,POET,FSE2015,JouleGuard} and one submitted publication
whose outcome is still pending as of this writing.