Computer Science Clinic Overview

Transcription

1 Harvey Mudd Coege Computer Science Department Computer Science Cinic Overview Robert M. Keer Professor & Director hmc.edu/cinic hmc.edu

2 History of Computer Science Cinic 1992: Computer Science Major Estabished 1993: Computer Science Cinic Estabished : 4 Projects (1 joint with Math, Bio, : 3 Projects : 3 Projects (1 joint with Math, Bio, Engrg.) : 7 Projects (1 joint with Engineering) : 5 Projects : 7 Projects : 9 Projects : 11 Projects + 1 summer project : 11 Projects (1 joint with Engineering) + 1 summer : 10 Projects

3 Cients New Cients: Green Media Tooshed Kofax Image Products LaserFiche Maguandia Studio Northrop Grumman Sandia Nationa Laboratories Repeat Cients: The Aerospace Corporation Auditude Boeing Teradyne

4 Project Areas n Network Management and Security Aerospace Hand-Hed Mobie Appication Boeing Artificia Inteigence/Pattern Recognition Auditude, Kofax, LaserFiche, Sandia Graphics/Animation Maguandia Studio Faut-Toerance Northrop Grumman Web Enterprise Appication Teradyne Database Appication Green Media Tooshed

5 Facuty Areas Zachary Dodds Mike Eringer Robotics, Vision Network Management, Security Ran Libeskind-Hadas Robert Keer Geoff Kuenning Meissa O NeiO Chris Stone Eizabeth Sweedyk Beinda Thom Distributed Computing Agorithms Languages, Parae Processing Networking and Operating Systems HCI, Distributed computing Programming Languages Graphics, Animation Machine Learning, Music

6 Components of the CS Major Principes of Computer Science Kerne Courses (6) Technica Eectives (3 minimum) CS Cinic (1 year) Cooquium (2 years) Research projects (optiona)

7 Kerne CS Courses Data Structures and Program Deveopment Computabiity and Logic Agorithms Computer Systems Software Deveopment Programming Languages

8 Computer Science Eectives Advanced Computer Architecture Advanced Operating Systems Artificia Inteigence Compier Design Computer Graphics Computer Networks Computer Vision Databases Neura Networks Parae and Rea-Time Computation Performance Anaysis Robotic Theory of Computation

9 What Typica Cinic Projects Entai Nuera produces switches that route teephone cas. Designed and buit software simuator for networks for agorithm testing, and Software for automaticay generate efficient routing tabes

10 Typica Projects (cont d) Optivus Technoogy deveops radiation treatment systems based on Proton beams. Deveoped a registration too for radiographic images to speed the patient aignment process whie enhancing accuracy.

11 Typica Projects (cont d) IBM Austin Research Center is interested in high-performance processors based on asynchronous ogic. Deveoped a suite of toos for verifying hardware transformations, incuding a VHDL compier and visuaization capabiities

12 Typica Projects (cont d) Octe Communications needed to transate a arge set of voice-mai prompts into many different anguages Deveoped an interactive too used by human transators that provides phrase structure and contextua information

13 Typica Projects (cont d) Microsoft wanted the abiity to dispay Sidewak web-sites on hand-hed devices. Deveoped dispay and down-oad software for severa popuar patforms (Piot, Windows CE)

14 Design Goas from Microsoft Project Provide the user as much information as possibe on the device without consuming unreasonabe resources Dispay the data in a way famiiar to the user Make the process of maintaining current information neary invisibe to the user

15 Project Impementation Overview Downoad data: Internet Linkage Parse data: Cient-Side HTML Parsing Transfer data: PC Dock to Handhed Computer Dispay data: User Interface Design

16 Deiverabes Sidewak for Piot 1.0 PC appication, conduit dynamic ink ibrary, Piot browser, instaer Sidewak for Windows CE 1.0 PC appication, instaer Source code and documentation

17 Concusion Computer Science Cinic projects address Technica Requirements Performance Requirements Ergonomic Requirements ranging from current practice to advanced research in a manner cost-effective to cients.

18 Unsoicited Assessment Letter I've received the fina report and read it this morning. I beieve the cinic team has done a most wonderfu job. The first probem that was tacked, finding a good VHDL compier for asynchronous design, turned out to be a ot more work than I had anticipated (having been ucky in trying things ony on the one compier that seems to work before I asked the team to ook at asynchronous VHDL). I think it is absoutey amazing that the team managed to overcome the probems with finding a good compier in such a short time, and sti carry out most of the (aggressive) goas that were set at the start of the project.

19 Letter, Continued The VHDLemu system that was deveoped is a very decisive way to dea with compier probems indeed, and that system by itsef woud have made for a great cinic end resut. I think it puts some very expensive academic and industry VHDL deveopment effort to shame,, by achieving higher functionaity and correctness rather than tons of whistes and bes on a fauty core. I am aso very impressed with the mechanized proof that deadock is avoided in the MIPS decomposition. I see this as a major achievement.. I know that the underying theory is difficut to communicate, and yet the team understood it so we they coud turn it into a too. I aso appreciate the competeness of that effort. My chaenge to the team had been to verify some transformations, and I had certainy not anticipated to see the compete decomposition anayzed. If you ook a itte in the forma methods and mechanized proof iterature, you wi see that this work is on the cutting edge!

20 Letter, Concuded Finay, the visuaization too very nicey ties the whoe thing together. Adding concurrency and knowing that you have not sacrificed correctness is one thing, but convincing yoursef that the added concurrency gives added performance is another, and a visuaization too is a great way to give insight into what has been achieved, or where the bottenecks are.

21 CS Cinic Sponsors Acce Technoogies, Inc. The Aerospace Corporation (5) Beckman Instruments GTE HRL Laboratories IBM (3) Jet Propusion Laboratories (2) Legato Systems, Inc. Microsoft Corporation (2) Nuera Communications Octe Communications Corp. (2) Optivus Technoogy, Inc. (4) Parace,, Inc. Rockwe Science Center Teradyne, Inc. WebTV Networks, Inc. Wordport,, Inc.