Csaba Andras Moritz

Random Info

[RandomInfo]

[Contact ]

[Projects]

[Students]

[Classes]

[Talks]

[Selected Publications ]

[All Publications ]

Csaba Andras Moritz
E-mail:
andras@ecs.umass.edu
Phone:
(413) 545-2442
E-fax:
(603) 697-0230

Secretary:
June Daehler
Phone:
(413)-545-3621

Address:
University of Massachusetts, Amherst,
Elect. and Comp. Eng.
Knowles Eng. Building,
309H, Amherst, MA 01003
Map

I'm an associate professor in the Electrical and Computer Engineering department at the University of Massachusetts Amherst, and visiting scientist at the Laboratory for Computer Science MIT. I am also teaching in the Computer Science department at UMASS. Before joining UMASS I was a research scientist at MIT (1997-2000). See my webpage at MIT here. I have earned my PhD in computer systems from the Royal Institute for Technology, Stockholm, Sweden. During my career I have been involved with companies for almost a decade as CTO, consultant, and founder of a distributed systems company. (More recently also in the Carl Bernadotte Asset Management Venture Capital Co in Stockholm, Sweden.). I have worked in 5 countries, several companies, and have been employed in 5 universities (this despite that my background is mostly industrial). My biggest contribution (I am joking) as an advisor was to convert my wife into an expert and enthusiastic software engineer in a leading software company after no interest whatsoever from her part (she has formal education in literature/arts).

My current research focus is on two areas: (1) The Cool-* project (read Cool Star), compiler-enabled power-aware architectures, (2) The Euphoria project, sel-evolving software/hardware using introspection.

I am building a research group in Software Systems & Architectures and I am actively looking for motivated, bright graduate students who are interested in building systems for these new projects. My projects span across several fields: architecture, compilers, software systems, and analytical modeling. Skills that help you succeed: excellent system building "hacking" in Java/C/C++, computer architecture, compilers, and distributed systems. I regularly receive emails from prospective students. My advise to you all: rather than telling me about how interesting my group's research is and how excited you are to join my group with an RA position, I would like you to focus on how would/can you contribute concretely to that research.

My older projects include Hot Pages, a software caching system for Raw microprocessors, and SUDS , a speculative execution system for Raw, in particular the design and implementation of the runtime system for speculative-execution . Other research include LoGPC , a model for estimating network contention on multiprocessors, SimpleFit or analytical modeling of Raw architectures, MPI performance evaluation and modeling (on Cray T3D, Convex Exemplar, Meiko and network of workstations), silicon compilation, and multiprocessor scheduling.

My general research interests include:

(compiler-enabled, reconfigurable, parallel, single-chip) computer architecture,
software systems (runtime systems, compilers),
modeling,
scalable Internet application infrastructure.

(click to enlarge picture)

Projects
check out the Software Systems & Architecture Lab for a more updated version.

FlexCache is a novel framework for a compiler enabled memory system. It has no hardware cache controller (can be combined with it though) and it uses architecture support to reduce the latency of the hit case by implementing special speculative memory instructions. A FlexCache design can provide unprecedented flexibility by supporting multiple individually controlled partitions, high associativity, and variable size cache blocks. We plan to target FlexCache primarily to embedded systems. Components include:
- automatically synthesize caches for reconfigurable architectures or FPGAs,
- to achieve higher predictability in caches for real-time systems,
- to obtain high caching efficiency in superscalars,
- and to reduce power consumption in memory systems.
- applications include: streaming applications, earthquake simulations

Euphoria is a framework for self-modifying computations in distributed systems. It focuses on scalable software infrastructure for global dynamic content management applications on cluster of workstations and multiprocessors. Scalability is automated and controlled with introspection. Euphoria implements it's scalable resource management technology based on:
- byte code level analysis and transformations,
- individualized content management,
- object cache store management,
- non-intrusive introspection that uses static compile-time information,
- analytical forecasting models,
- web caching for applications,
- wireless - networked client integration HTML/XML-WAP
- real-time JVM for resource constrained devices
From the outside it aims to be compatible with application server and client software technologies such as defined in J2EE, J2ME, CLDC, WAP, KVM specifications and have support for both traditional web based and wireless clients.
MIT-RAW , software virtualization and memory microoptimization techniques
- Hot Pages
- Compiler Managed Memory

Fine Grained Synchronization in Multiprocessors
- integrating cache coherency with fine grained synchronization in SMP/NUMA/single-chip multiprocessors

Integrated Analytical and Compiler Generated Simulation of Billion Transistor Architectures
- Integrated Framework for Cost Performance Evaluation of Architectures with 1000s of processing units
- SimpleFit - analytical modeling of Raw and FPGAs

Courses I am teaching

ECE department: ECE697E Software Exposed Architectures (Fall 2000)
CS department: CS377: Operating Systems (Spring 2001),
Reading group on Internet Middleware (Spring 2001) sign up here
Reading group on Advanced Compiler Techniques (Fall 2001) sign up here
At ECE: ECE669, Parallel Computer Architecture (Fall 2001)

Students in my group

check out the Software Systems & Architecture Lab

Recent Selected Publications
(More publications )

Csaba Andras Moritz, Matthew Frank, Saman Amarisinghe
FlexCache: A Framework for Flexible Compiler Generated Data Caching
Appears in IRAM00, 2-nd Intelligent Memory Workshop, Cambridge, Nov 2000

Csaba Andras Moritz, Matthew Frank, Walter Lee,and Saman Amarasinghe
Hot Pages: Software Caching for Raw Microprocessors ,
Submitted for publication to ISCA-27, Nov 1999.
[Compressed Postscript] [Postscript] [PDF]
Csaba Andras Moritz, Matthew Frank, Walter Lee,and Saman Amarasinghe
Hot Pages: Software Caching for Raw Microprocessors ,
MIT-LCS Technical Memo LCS-TM-599, Aug 1999.
[Compressed Postscript] [Postscript] [PDF]
Jonathan Babb, Martin Rinard, Csaba Andras Moritz, Walter Lee, Matthew Frank, Rajeev Barua, and Saman Amarasinghe
Parallelizing Applications into Silicon,
The 7th Annual IEEE Symposium on Field-Programmable Custom Computing Machines
FCCM'99, Napa, California, April 1999.
[Compressed Postscript] [Compressed Postscript, Tech Report]
Matthew Frank, Csaba Andras Moritz, Benjamin Greenwald, Saman Amarasinghe and Anant Agarwal
SUDS: Primitive Mechanisms for Memory Dependence Speculation,
Technical Report. LCS-TM-591, October 1998.
[Compressed Postscript] [Postscript] [PDF]
Csaba Andras Moritz, M. Frank.
LoGPC: Modeling Network Contention in Message-Passing Programs.
ACM Joint International Conference on Measurement and Modeling of Computer Systems,ACM SIGMETRICS/PERFORMANCE 98 Wisconsin Madison, June 1998,
,also in ACM Performance Evaluation Review Special Issue Vol 26 no 1.
[Compressed Postscript] [Postscript] [PDF]
Csaba Andras Moritz, D. Yeung, A. Agarwal.
Exploring Cost-Performance Optimal Designs of Raw Microprocessors.
The 6th Annual IEEE Symposium on Field-Programmable Custom Computing Machines,
FCCM'98, Napa, California, April 1998.
new updated version will appear in IEEE Transactions on Parallel and Distributed Systems,
[Compressed Postscript] [Postscript] [PDF]
Csaba Andras Moritz.Towards Optimal Resource Utilization in Parallel Computer Systems.
PhD thesis (contains several earlier papers)
Royal Institute of Technology, Stockholm, Sweden.
[Compressed Postscript]

Recent Talks

FlexCache: A Farmework for Compiler Generated Data Caching, 2-nd Intelligent Memory Workshop, Cambridge, MA, Nov 2000,(request it here)
HotPages: Software Caching for Raw Microprocessors (request it here)
LoGPC: Modeling Network Contention in Message-Passing Programs, ACM Joint International Conference on Measurement and Modeling of Computer Systems,
SIGMETRICS/PERFORMANCE 98, June 1998,Wisconsin Madison, June 1998.
[Compressed Postscript]. [Postscript]. [PDF].
Exploring Cost-Performance Optimal Designs of Raw Microprocessors. The 6th Annual IEEE Symposium on Field-Programmable Custom Computing Machines
FCCM'98., April 1998, Napa, California
[Compressed Postscript]. [Postscript]. [PDF].
Adaptive Software Virtual Memory for Raw. Laboratory for Computer Science, Raw group, MIT, Jan 1999, Cambridge
[PDF].
Design and Implementation of Software Caching for Raw. IBM Research Lab Austin, May 1999,
[HTML].

Contact Info
E-mail: andras@ecs.umass.edu
Phone: (413) 545 2442
E-fax: (603) 697-0230

     Csaba Andras Moritz
     University of Massachusetts Amherst
     Knowles Building
     Amherst, MA 01003

This page last updated on January, 2001.andras@ecs.umass.edu