Alexander M. Merritt
Ph.D. Pre-candidate, Computer Science
College of Computing
Georgia Institute of Technology
I am a third-year Computer Science Ph.D. student in the College of Computing at the Georgia
Institute of Technology, advised by Dr. Karsten
Schwan. I completed my B.S. in Computer Science with high honors and a minor in German from the
Rochester Institute of Technology in May 2009. I spent three years
growing up in Tuebingen, Germany and a little over a decade in Highland Park, New Jersey.
My research interests lie within systems software and operating systems, targeting modern
many-core, accelerator and high-performance architectures. I am driven to understand and improve the
intersection of these layers, allowing applications to best exploit these systems while considering
portability, programmability and efficiency of use.
Specific examples include
- Providing virtual platforms of GPGPUs for enabling applications to scale despite physical
limitations of compute nodes to contain arbitrary compositions of GPGPUs
- Examining operating system support for efficient use of NUMA systems for threaded, SMP-based
- Investigating memory consistency models within systems software on highly many-core
architectures (Intel's SCC) in the absence of hardware-enabled cache coherence protocols
Current Research Projects
- Shadowfax --- GPGPUs have proven to be advantageous for increasing application
scalability both in the HPC and enter- prise domains. This has resulted in an increase in the array
of programming languages and range of physical compute capabilities of current hardware. Yet
applications’ scalability and portability remain limited with respect to both their degree of
customization and the physical limitations of compute nodes to contain any number and composition of
devices. This research defines the notion of a GPGPU assembly for CUDA ap- plications resident in
Xen virtual machines as well as non-virtualized nodes in high-performance clusters, presenting to
applications a set of GPGPUs as locally-available devices to best match their needs, easing
programmability and portability. We characterize workloads to best match them with available GPGPUs
and employ techniques such as dynamic resource monitoring of CPU, network and memory usage as well
as instrumenting binary CUDA kernels to obtain GPGPU-level usage to enable higher-level mapping
poli- cies. This aims to provide a framework for more intelligent management of GPU clusters (e.g.
providing performance isolation), examining global scheduling policies, admission control and
dynamic retargeting of execution streams.
- Alexander Merritt, Vishakha Gupta, Abhishek Verma, Ada Gavrilovska, Karsten Schwan;
"Shadowfax: Scaling in Heterogeneous Cluster Systems via GPGPU Assemblies";
VTDC'11, San Jose, CA.
- John R. Lange, Kevin Pedretti, Peter Dinda, Chang Bee, Patrick Bridges, Philip Soltero,
Alexander Merritt; "Minimal-overhead Virtualization of a Large Scale Supercomputer";
VEE'11, Newport Beach, CA.
- Alexander Merritt, Kevin Pedretti, "Techniques for Managing Data Distribution in NUMA
Systems", Sandia National Labs Computer Science Research Institute Technical Report (2010),
- Alexander Merritt, Kevin Pedretti, "Techniques for Managing Data Distribution in
NUMA Systems", SC'10, New Orleans, LA, November 2010.
- President's Fellowship, Georgia Institute of Technology.
- Dean's List, Golisano College of Computing and Information Sciences, Rochester Institute of Technology.
I live in Atlanta, GA. My desk is located in the Klaus Advanced Computing Building, room 3201.
last modified: December 11 2011.