Date: Monday, 25-Nov-96 21:50:31 GMTServer: NCSA/1.1MIME-version: 1.0Content-type: text/htmlLast-modified: Wednesday, 10-Jan-96 21:41:33 GMT<CENTER><title>Patricia J. Teller</title><br><h1> Patricia J. Teller, Assistant Professor </h1> <RIGHT><!WA0><img src="http://www.cs.nmsu.edu/~teller/little_teller.gif"></RIGHT><br><!WA1><A href="http://www.nmsu.edu"> New Mexico State University </A><br><!WA2><A href="http://www.cs.nmsu.edu">Department of Computer Science</A><br>Science Hall, Room 123<br>Stewart and Sweet Streets<br>Las Cruces, New Mexico 88003-8001<br><br><h4>teller@cs.nmsu.edu<br>http://www.cs.nmsu.edu/~teller<br>505-646-6243<br></h4></CENTER><p> Pat received her B.A., M.S., and Ph.D. from <!WA3><A href="http://www.nyu.edu">New York University</A>.  Her Ph.D.advisers were Drs. <!WA4><A href="http://www.cs.nyu.edu/cs/faculty/gottlieb/index.html"> Allan Gottlieb</A> and <!WA5><A href="http://www.cs.nyu.edu/cs/faculty/grishman/index.html">Ralph Grishman</A>.  She is a member ofPhi Beta Kappa and an honorary member of the Golden Key National Honor Society.  Before joining NMSU, Teller was a visiting researcher at IBMT. J. Watson Research Center.  Dr. Teller's main research interests are parallel and distributed computing, computer architecture, operating systems,and performance evaluation.  These research interests are the main focusof her Parallel Research Group at NMSU.  Dr. Teller's research has beenfunded by Sandia National Laboratories, the U. S. Army Research Office, andLos Alamos National Laboratory. Teller is a reviewer for major conferencesand journals and has been on the program committee of several major conferences.  She has been invited to speak at various workshops, conferences,universities, and industrial centers.<p>The Performance Research Group (PRG) at NMSU includes five of Pat's Ph.D.candidates, <!WA6><A href="http://www.cs.nmsu.edu/~kbisset">Keith Bisset</A>,  <!WA7><A href="http://www.nmsu.edu/~scooper">Shaun Cooper</A>,  <!WA8><A href="http://www.cs.nmsu.edu/~hsalgado">Hugo Molina-Salgado</A>,  <!WA9><A href="http://www.cs.nmsu.edu/~roliver">Richard Oliver</A>, and Qidong Xu (now working at IBM)); two of her Master's candidates,  <!WA10><A href="http://www.cs.nmsu.edu/~wmcgrego">Ward McGregor</A>) and <!WA11><A href="http://www.cs.nmsu.edu/~rsinclai">Rob Sinclair</A>); andseveral other students including Ph.D. candidates <!WA12><A href="http://www.cs.nmsu.edu/~jechen">Jen-Long Chen</A>); and <!WA13><A href="http://www.cs.nmsu.edu/~hdp">Heather Pfeiffer</A>); andMaster's candidate  <!WA14><A href="http://www.cs.nmsu.edu/~fbassett">Federico Bassetti</A>); andand Bachelor's candidate, Praveen Mamnani.<p><h2> <!WA15><A HREF="http://www.cs.nmsu.edu/~teller/myvita.ps"> VITA </A> </h2><h2> PUBLICATIONS (this section is under construction)</h2><!WA16><A href="http://www.cs.nmsu.edu/~teller/publications/tech.reports/unscheduled.traces.ps"> Unscheduled traces and shared-memory multiprocessor simulation,"Patricia J. Teller,NMSU Technical Report NMSU-CSTR-9426, 1995.</A><p>Abstract.  Trace-driven simulation of shared-memory architectures can be as accurate as execution- or program-driven simulations for a large class of parallel programs. Traces for programs in this class can be scheduled or unscheduled. A scheduled trace is a set of traces generated by running aparallel program on a real or simulated MP, while an unscheduled trace is an augmented trace generated by running a parallel program on a uniprocessor. Scheduled traces can be used to drive simulations of only MP systems with the same number of processors as the MP that generated the trace and are limited to the same static task-to-processor scheduling as the MP that generated the trace. In contrast, unscheduled traces offer much more flexibility. In this paper we show how unscheduled traces can be used to simulate dynamic task-to-processor scheduling, MP systems of different sizes, and multiprogramming. In addition, we show how an unscheduled trace allows the task granularity of a traced program to be varied from simulation to simulation and how operatingsystem references can be inserted dynamically during simulation.<p><h2> <!WA17><A href="http://www.cs.nmsu.edu/~teller/CS579.Fall95/index.html"> CS579 Simulation of Computer Architectures </A> </h2>.CS579 is a special topics course for graduate students.  The course willcover research in the area of simulation of computer architectures, bothuniprocessor and multiprocessors architectures.  You can examine the syllabus of the course, lecture notes, and papers that are being analyzed by the students, among other things.<p><h2> CS273 Course Materials</h2>  <p>CS273 is a course in Machine Organization and Assembly Language.  This course is described in "Mobile robots teach machine-level programming",which will appear in the proceedings of Supercomputing '95.  Course materials,including transparencies and a MC68HC11 simulator with a graphical user interfacewill be available at this WWW site during or before December 1995.<p>The following lecture notes are available for your perusal:<UL><LI><!WA18><A HREF="http://www.cs.nmsu.edu/~teller/CS273/lecture.notes/FirstLectures/CS273Fall95.FirstLectures.ps">Motivational lectures, which include ... </A>;<LI><!WA19><A HREF="http://www.cs.nmsu.edu/~teller/CS273/lecture.notes/ProgrammingConstructs/CS273Fall95.ProgrammingConstructs.ps">Templates for converting high-levellanguage programming constructs into low-level (assembly) language constructs.</A>;<LI><!WA20><A HREF="http://www.cs.nmsu.edu/~teller/CS273/lecture.notes/Stacks/CS273Fall95.Stacks.ps">Lecture on implementation of stacks.</A>;</UL>Click <!WA21><A HREF="http://www.cs.nmsu.edu/~teller/CS273/reading"> here </A> tosee a description of the reading that has been assigned in the course.<p>For the fall 1995, click <!WA22><A HREF="http://www.cs.nmsu.edu/~teller/CS273/assignments"> here </A> to access the assignments that have beenassigned thus far.<p>More to come .....