Lecture on February 06, 2002 0:00:00 - 0:06:20 Course Administration 0:06:20 - 0:07:43 Course Administration: the computer we'll use 0:07:43 - 0:11:10 Course Administration: schedule of projects 0:11:10 - 0:12:15 Course Administration: class topics and calendar 0:12:15 - 0:19:15 Parallel Architectures 0:19:15 - 0:30:00 Parallel Architectures: architecture details 0:30:00 - 0:32:00 Parallel Architectures: more details 0:32:00 - 0:42:45 Parallel Architectures: more details: diagrams of parallel architectures 0:42:45 - 0:47:18 Parallel Architectures: more details: SMPs 0:47:18 - 0:51:25 Parallel Architectures: more details, speeds 0:51:25 - 0:53:00 Parallel Architectures: Moore's law 0:53:00 - 0:59:00 Parallel Architectures: pictures of supercomputers 0:59:00 - 1:04:15 Applications 1:04:15 - 1:07:00 Applications: goals of parallel computing 1:07:00 - 1:09:00 Applications: embarrassingly parallel applications 1:09:00 - 1:15:00 Special Approaches 1:15:00 - 1:23:00 Beowulf Lecture on February 11, 2002 0:00:00 - 0:05:20 Start, course administration 0:05:20 - 0:08:40 Lattice QCD 0:08:40 - 0:12:10 Lattice QCD: what is QCD and why is it important? 0:12:10 - 0:20:15 Lattice QCD: lattice QCD computations 0:20:15 - 0:28:30 Lattice QCD: clusters for LQCD computation 0:28:30 - 0:29:30 Lattice QCD: conclusion 0:29:30 - 0:36:00 Weather and climate change; weather: history of parallel computing 0:36:00 - 0:40:40 Weather: weather: parallel computing now and future 0:40:40 - 0:47:00 Weather: climate change 0:47:00 - 1:00:30 Weather: activities at MIT 1:00:30 - 1:09:30 Weather: questions 1:09:30 - 1:22:00 Bad parallel algorithms Lecture on February 13, 2002 0:00:00 - 0:01:20 Start, lecture outline 0:01:20 - 0:12:00 First homework assignment 0:12:00 - 0:12:40 Connecting to the class machine, beowulf.lcs.mit.edu 0:12:40 - 0:23:30 OpenMP introduction 0:23:30 - 0:55:20 OpenMP details and example 0:55:20 - 0:59:30 OpenMP running on our beowulf 0:59:30 - 1:22:52 MPI Lecture on February 19, 2002 0:07:00 - 0:15:00 Start 0:15:00 - 0:26:00 Parallel Prefix Algorithm 0:26:00 - 0:42:00 Associative Operations 0:42:00 - 0:47:00 The "Myth" of log n 0:47:00 - 0:54:00 Prefix Operation Segmented 0:54:00 - 1:01:00 Fortran 1:01:00 - 1:10:00 Parallel Prefix Variations 1:10:00 - 1:24:00 PRAM Lecture on February 20, 2002 0:07:00 - 0:08:00 Start(Parallel Computer Architecture I) 0:08:00 - 0:14:00 Latency/Bandwidth 0:14:00 - 0:23:00 Latency: the details 0:23:00 - 0:31:00 Node Architecutre 0:31:00 - 0:36:00 Bus as an interconnect network 0:36:00 - 0:40:00 Asic White at Lawrence Livermore 0:40:00 - 0:48:00 Cross Bar 0:48:00 - 0:55:00 Cache/Cache Coherence 0:55:00 - 1:09:00 CM-2 1:09:00 - 1:20:00 Hypercube 1:20:00 - 1:23:00 Routing 1:23:00 - 1:25:00 Paradiso Cafe Problem Lecture on February 25, 2002 0:07:00 - 0:14:00 Start 0:14:00 - 0:27:00 Linear Algebra Libraries 0:27:00 - 0:29:00 History, Top 500 ... 0:29:00 - 0:39:00 Optimizing Computation and Memoery Use 0:39:00 - 0:43:00 BLAS 0:43:00 - 0:47:00 Self Adapting Numerical Software 0:47:00 - 0:53:00 Software Generation Strategy 0:53:00 - 1:01:00 Gaussian Elimination 1:01:00 - 1:04:00 Distributed and Parallel Systems 1:04:00 - 1:07:00 Three Basic Linear Algebra Problems 1:07:00 - 1:08:00 Results for Parallel Implementation on Intel Delta 1:08:00 - 1:26:00 More on Gaussian Elimination (Reorganization) 1:26:00 - 1:28:00 ScaLAPACK and MATLAB*P Lecture on February 27, 2002 0:07:00 - 0:08:00 Linear Algebra: Start 0:08:00 - 0:11:00 Linear Algebra: Fundamental Triangle 0:11:00 - 0:12:00 Linear Algebra: Algorithm & Architecture 0:12:00 - 0:16:00 Linear Algebra: Architecture ... 0:16:00 - 0:20:00 Dense Linear Algebra 0:20:00 - 0:22:00 Linear Algebra: Basic Algorithm Change 0:22:00 - 0:29:00 Linear Algebra: FMA Instruction 0:29:00 - 0:30:00 Blocking 0:30:00 - 0:34:00 Linear Algebra: Recursion 0:34:00 - 0:35:00 Block Column Major Order 0:35:00 - 0:39:00 Linear Algebra: Square Block ... 0:39:00 - 0:42:00 Blocked Mat-Mult is Optimal 0:42:00 - 0:44:00 Linear Algebra: Matrix Multiplication is Pervasive 0:44:00 - 0:48:00 Linear Algebra: Recursion 0:48:00 - 0:49:00 Linear Algebra: Standard Full Format 0:49:00 - 0:51:00 Linear Algebra: Block Hybrid Full Format 0:51:00 - 0:55:00 Linear Algebra: Blocked Based Algorithms Via LAPACK 0:55:00 - 0:57:00 Linear Algebra: Concise Algorithms Emerge 0:57:00 - 1:01:00 Linear Algebra: Tree Diagram of Cholesky Algorithm 1:01:00 - 1:04:00 Linear Algebra: Challenge of Machine Independent Design of Dense Linear Algebra Codes via the BLAS 1:04:00 - 1:06:00 Linear Algebra: Can we exploit this general relationship? 1:06:00 - 1:07:00 Recursive Data Format 1:07:00 - 1:11:00 Linear Algebra: Dimension Theory 1:11:00 - 1:12:00 Linear Algebra: Changes 1:12:00 - 1:13:00 Linear Algebra: New LAPACK Type Routine 1:13:00 - 1:20:00 Linear Algebra: Answering questions from audience Lecture on March 04, 2002 0:00:00 - 0:01:00 Start: course administration 0:01:00 - 0:15:00 Matlab demo 0:15:00 - 0:32:00 Matlab Operator overloading; sparse matrices; vectorization 0:32:00 - 0:50:00 Matlab*P overview 0:50:00 - 0:57:00 Matlab*P demo 0:57:00 - 1:02:00 FEMLAB overview 1:02:00 - 1:17:00 FEMLAB demo 1:17:00 - 1:25:00 FEMLAB: how it works 1:25:00 - 1:26:00 Project comments Lecture on March 06, 2002 0:08:00 - 0:09:00 Start 0:09:00 - 0:30:00 MATLAB*P Demo 0:30:00 - 0:34:00 N-Body Problem 0:34:00 - 0:40:00 N-Body Problem: What is the Computation? 0:40:00 - 0:42:00 N-Body Problem: O(n^2)? Right? 0:42:00 - 0:46:00 N-Body Problem: Variations 0:46:00 - 0:50:00 N-Body Problem: O(n^2) VS O(nlog(n)) 0:50:00 - 0:53:00 N-Body Problem: nlog(n) Type of Computation 0:53:00 - 0:57:00 Data Structure: Quad-tree 0:57:00 - 1:00:00 Data Structure: Qct-tree 1:00:00 - 1:08:00 Barnes-Hut 1:08:00 - 1:11:00 Multipole (in 1D, constant potentials) 1:11:00 - 1:15:00 Multipole (in 1D, potential= quadratic polynomials) 1:15:00 - 1:17:00 Multipole (global coordinates vs local coordinates) 1:17:00 - 1:26:00 Multipole (Vi(x)=qi/(x-xi)) 1:26:00 - 1:28:00 Multipole Series Lecture on March 11, 2002 0:00:00 - 0:02:00 Start: projects and homework schedule 0:02:00 - 0:07:00 Fast multipole: quick review; adding functions 0:07:00 - 0:12:30 Fast multipole: Analogy with finite precision arithmetic 0:12:30 - 0:21:00 Exclusion sum as matrix/vector multiplication 0:21:00 - 0:25:20 Exclusion sum as matrix decomposition 0:25:20 - 0:29:15 Multipole as matrix decomposition 0:29:15 - 0:32:00 Representing functions: Taylor series 0:32:00 - 0:45:30 Interpolating polynomials 0:45:30 - 0:52:00 Matlab: Symbolic example 0:52:00 - 1:04:30 Multipole series (as Taylor series in 1/x) 1:04:30 - 1:10:00 Virtual charges 1:10:00 - 1:17:45 Fitting a polynomial to the example 1:17:45 - 1:18:45 Adding representations of functions 1:18:45 - 1:20:10 Summary of the whole fast multipole algorithm Lecture on March 13, 2002 0:00:00 - 0:01:43 Start: setting up A/V 0:01:43 - 0:02:25 Tape running. Project proposal status 0:02:25 - 0:10:00 Multipole continued: object-oriented Matlab; exclude2d 0:10:00 - 0:23:50 Taylor series in Matlab by operator overloading 0:23:50 - 0:25:00 Exclusion sum on Taylor series 0:25:00 - 0:33:15 Multipole series in Matlab 0:33:15 - 0:41:00 Addition algorithms using binomial coefficients 0:41:00 - 0:45:00 Code for multipole addition; the Pascal matrix; example 0:45:00 - 0:50:00 Multipole objects with symbolic entries (!) 0:50:00 - 0:55:30 Multipole: summing up the multipole sum algorithm 0:55:30 - 1:23:00 Multipole: Overall summary, questions 1:23:00 - 1:25:10 Multipole: History and other applications Lecture on March 18, 2002 0:00:00 - 0:01:07 Start; project administration 0:01:07 - 0:04:30 Parallel architecture II; caches 0:04:30 - 0:16:00 Cache coherence and how it's enforced 0:16:00 - 0:27:30 Cache coherence: Write invalidate vs write update protocols 0:27:30 - 0:32:40 Cache coherence: Write invalidate with write-back cache 0:32:40 - 0:39:20 Distributed shared memory 0:39:20 - 0:45:40 The future of cache (according to RAW) 0:45:40 - 0:51:30 A little about the RAW architecture 0:51:30 - 1:00:00 More fun with multipole 1:00:00 - 1:09:30 Experiments with multipole accuracy; sources of rounding error 1:09:30 - 1:21:56 Questions about Beowulf Lecture on March 20, 2002 0:00:00 - 0:02:00 Start; administration and announcements 0:02:00 - 0:03:25 Beowulf: Building 0:03:25 - 0:07:00 Beowulf: History 0:07:00 - 0:10:00 Beowulf: List of clusters on the web 0:10:00 - 0:17:30 Beowulf: definition, motivation 0:17:30 - 0:37:45 Beowulf: Hardware options 0:37:45 - 0:48:08 Beowulf: Experience with hardware 0:48:08 - 0:57:15 Beowulf: Software options 0:57:15 - 1:00:50 Beowulf: Experience with software 1:00:50 - 1:05:20 Beowulf: Recipe (what really happened) 1:05:20 - 1:08:00 Beowulf: Exercise: Design a $30,000 Beowulf 1:08:00 - 1:20:50 Questions and discussion Lecture on April 01, 2002 0:00:00 - 0:04:30 Start, Overview 0:04:30 - 0:10:20 MATLAB Demo 0:10:20 - 0:20:00 Sparse Matrices in Real Life 0:20:00 - 0:21:00 MATLAB Matrices: Design Principles 0:21:00 - 0:22:30 Data Structures 0:22:30 - 0:34:00 Algorithms (Ax=b) 0:34:00 - 0:35:20 Solving Linear Equations: x=A\b 0:35:20 - 0:41:30 Graphs and Sparse Matrices: Cholesky Factorization 0:41:30 - 0:46:00 Elimination Tree 0:46:00 - 1:01:00 MATLAB Demo: Sparse Matrices and Graphs 1:01:00 - 1:06:10 Fill Reducing Matrix Permutations 1:06:10 - 1:12:15 Matching and Block Triangular Form 1:12:15 - 1:16:46 Complexity of Direct Methods 1:16:46 - 1:25:00 The Landscape of Sparse Ax=b Solvers Lecture on April 03, 2002 0:00:00 - 0:10:00 Start, related webpages 0:10:00 - 0:15:00 Direct Methods 0:15:00 - 0:17:00 GEPP: Gaussian elimination w/ partial pivoting 0:17:00 - 0:21:30 Symmetric Positive Definite: A=R'R 0:21:30 - 0:26:00 Symbolic Gaussian Elimination 0:26:00 - 0:27:30 Sparse Triangular Solve 0:27:30 - 0:33:00 Left-looking Column LU Factorization 0:33:00 - 0:35:30 Symmetric Supernodes 0:35:30 - 0:38:40 Nonsymmetric Supernodes 0:38:40 - 0:40:50 Sequential SuperLU 0:40:50 - 0:44:40 Column Elimination Tree 0:44:40 - 0:47:00 Shared Memory SuperLU 0:47:00 - 0:48:40 Column Preordering for Sparsity 0:48:40 - 0:53:30 SuperLU dist: GE with static pivoting 0:53:30 - 0:58:00 Row permutation for heavy diagonal 0:58:00 - 1:05:30 Iterative refinement to improve solution 1:05:30 - 1:08:00 Question: preordering for static pivoting 1:08:00 - 1:15:30 Symmetric-pattern multifrontal factorization 1:15:30 - 1:18:40 MUMPS: distributed memory multifrontal 1:18:40 - 1:24:15 Remark on (nonsymmetric) direct methods Lecture on April 08, 2002 0:00:00 - 0:05:00 Start, Super LU-dist: iterative refinement 0:05:00 - 0:09:45 Convergence analysis of iterative refinement 0:09:45 - 0:14:45 The Landscape of Sparse Ax=b Solvers 0:14:45 - 0:22:00 Conjugate gradient iteration 0:22:00 - 0:31:15 Conjugate gradient: Krylov subspaces 0:31:15 - 0:37:15 Conjugate gradient: Convergence 0:37:15 - 0:47:15 Matlab demo 0:47:15 - 0:59:30 Conjugate gradient: Parallel implementation 0:59:30 - 1:07:00 Preconditioners 1:07:00 - 1:18:30 Incomplete Cholesky factorization 1:18:30 - 1:20:30 Sparse approximation 1:20:30 - 1:21:30 Support graph preconditioners: example 1:21:30 - 1:22:00 Multigrid 1:22:00 - 1:25:00 Complexity of direct methods Lecture on April 10, 2002 0:00:00 - 0:03:30 Start, introducing guest speaker, overview 0:03:30 - 0:09:30 Outlines: Embeded Stream Processing 0:09:30 - 0:13:45 Parallel Pipeline 0:13:45 - 0:16:30 Filtering 0:16:30 - 0:19:30 Beamforming and Detection 0:19:30 - 0:21:00 Types of Parallelism 0:21:00 - 0:28:00 Processing Algorithms: FIR overview 0:28:00 - 0:30:00 Processing Algorithms: Beamforming 0:30:00 - 0:33:00 Processing Algorithms: Detection 0:33:00 - 0:38:45 Parallelism Latency and Throughput 0:38:45 - 0:39:15 System Analysis: System Graph 0:39:15 - 0:47:15 System Analysis: Channel Space -> Beam Space 0:47:15 - 0:49:30 System Analysis: Dynamic Load Balancing 0:49:30 - 1:06:00 Software Framework 1:06:00 - 1:13:30 C++ Expression Templates and PETE 1:13:30 - 1:15:30 Performance Results 1:15:30 - 1:19:30 Matlab MPI 1:19:30 - 1:21:00 High Productivity Lauguage Experiments 1:21:00 - 1:25:00 Basic Msg Passing Lecture on April 17, 2002 0:00:00 - 0:01:50 Domain Decomposition (DD) 0:01:50 - 0:10:00 DD: Overlapping case of DD, example in FEMLAB 0:10:00 - 0:19:30 DD: Summary of overlapping DD 0:19:30 - 0:21:45 DD: History: why did Schwarz do this in 1870? 0:21:45 - 0:30:00 DD: Nonoverlapping DD, example in FEMLAB 0:30:00 - 0:39:00 DD: Normal derivatives on the boundary 0:39:00 - 0:40:00 DD: Summary of overlapping and nonoverlapping DD 0:40:00 - 0:42:45 Award ceremony 0:42:45 - 0:46:00 DD: Discretization 0:46:00 - 0:52:30 DD: The overlapping case: computational issues 0:52:30 - 1:00:40 DD: Jacobi and Gauss-Seidel iterations 1:00:40 - 1:06:00 DD: Overlappping DD as block Jacobi or block Gauss-Seidel 1:06:00 - 1:11:07 DD: Linear algebra formulation of Jacobi and Gauss-Seidel 1:11:07 - 1:20:00 DD: The nonoverlapping case: computational issues 1:20:00 - 1:23:45 DD: Solving the Schur complement system iteratively Lecture on April 22, 2002 0:00:00 - 0:03:30 Start, News on the Japanese fastest computer 0:03:30 - 0:05:30 Partitioning: Special Partitioning: One way to slice a problem in half 0:05:30 - 0:12:00 Partitioning: Laplacian of a graph 0:12:00 - 0:18:15 Partitioning: Edge-Node Incidence Matrix 0:18:15 - 0:27:30 Partitioning: Spectral Partitioning 0:27:30 - 0:39:15 Partitioning: Spectral Partitioning: Solve as an eigenvalue problem 0:39:15 - 0:46:30 Partitioning: Geometric Methods 0:46:30 - 0:50:00 Partitioning: Edge separator and Vertex Separator 0:50:00 - 0:52:45 Partitioning: Theory VS. Practice 0:52:45 - 0:59:00 Partitioning: Need Theoretical Class of Good Graphs 0:59:00 - 1:00:50 Partitioning: Geometric Separator Theorem 1:00:50 - 1:08:00 Partitioning: The Algorithm (Step 1 through 6) 1:08:00 - 1:18:45 Partitioning: Demo again and Radon Point 1:18:45 - 1:20:00 Partitioning: A few tricks 1:20:00 - 1:20:45 Partitioning: ParMETIS Lecture on April 24, 2002 0:00:00 - 0:12:00 Start, The fastest machine in Japan revisit. 0:12:00 - 0:14:45 SVMs 0:14:45 - 0:21:20 Supervised Learning 0:21:20 - 0:29:00 Linear Classification 0:29:00 - 0:30:30 The Optimization Problem 0:30:30 - 0:37:30 Non-separable Training Sets 0:37:30 - 0:44:30 The Dual Problem 0:44:30 - 0:50:00 Solving the Dual Problem 0:50:00 - 0:54:45 FMSvm Demo 0:54:45 - 1:12:00 MATLAB Example 1:12:00 - 1:20:00 Structural Risk Minimization Lecture on April 29, 2002 0:00:00 - 0:01:15 FFT: Start, topic: Fast Fourier Transform (FFT) 0:01:15 - 0:05:15 FFT: Definition of discrete Fourier transform 0:05:15 - 0:09:25 FFT: Pictures of FFTs 0:09:25 - 0:23:52 FFT: Example of phone tones 0:23:52 - 0:26:57 FFT: The FFT algorithm 0:26:57 - 0:29:30 FFT: Unshuffle 0:29:30 - 0:31:20 FFT: The matrix recurrence for Fn 0:31:20 - 0:37:00 FFT: Recursive form of the algorithm 0:37:00 - 0:44:45 FFT: Unwrapping the recurrence: bit reversal 0:44:45 - 0:47:45 FFT: Books on the FFT 0:47:45 - 0:53:00 FFT: Performance issues: the butterfly 0:53:00 - 0:57:20 FFT: Parallel issues: communication 0:57:20 - 1:05:30 FFT: notation: putting bars on digits, hypercube FFT 1:05:30 - 1:11:00 FFT: Back to parallel communication issues 1:11:00 - 1:21:07 FFT: Detailed look at a parallel FFT of size 32 on 4 processors 1:21:07 - 1:22:30 FFT: FFTW preview: fastest FFT in the West Lecture on May 01, 2002 0:00:00 - 0:00:30 Start 0:00:30 - 0:02:11 Cleve Moler introduction 0:02:11 - 0:04:30 History of Matlab 0:04:30 - 0:08:00 Fortran Matlab 0:08:00 - 0:17:00 Commercial Matlab 0:17:00 - 0:28:45 History of parallel computing 0:28:45 - 0:30:45 Amdahl's law 0:30:45 - 0:40:45 Why is it so hard to program parallel computers? 0:40:45 - 0:47:00 Cornell's Multi-Matlab 0:47:00 - 0:47:45 Nabeel Azar introduction 0:47:45 - 0:49:00 A first look at MultiMatlab 0:49:00 - 0:53:00 What is (and isn't) MultiMatlab? Why now? 0:53:00 - 0:56:26 Applications 0:56:26 - 0:57:24 Programming patterns 0:57:24 - 1:02:18 Single processor, multiple data (SPMD) style 1:02:18 - 1:07:10 Master/slave style 1:07:10 - 1:30:30 Discussion of implementation considerations Lecture on May 06, 2002 0:00:00 - 0:02:10 Start, course administration 0:02:10 - 0:03:10 FFTW: Introduction: Steve Johnson, Condensed Matter Physics, MIT 0:03:10 - 0:06:00 FFTW: Steve Johnson: FFTW, the "Fastest Fourier Transform in the West" 0:06:00 - 0:13:12 FFTW: Performance of FFTW 0:13:12 - 0:15:38 FFTW: Why FFTW is fast 0:15:38 - 0:19:20 FFTW: User's view of FFTW 0:19:20 - 0:19:36 FFTW: Outline of rest of talk 0:19:36 - 0:23:00 FFTW: The executor 0:23:00 - 0:34:50 FFTW: Cooley-Tukey FFT algorithm 0:34:50 - 0:37:25 FFTW: What a plan looks like 0:37:25 - 0:39:58 FFTW: Explicit recursion and out-of-cache FFTs 0:39:58 - 0:43:20 FFTW: Cache-oblivious FFT algorithms 0:43:20 - 0:47:15 FFTW: Vector recursion 0:47:15 - 0:49:00 FFTW: The planner 0:49:00 - 0:56:30 FFTW: How the planner works 0:56:30 - 0:59:45 FFTW: Why use adaptive programs? 0:59:45 - 1:01:05 FFTW: Self-optimization is easy 1:01:05 - 1:02:38 FFTW: The generator, genfft 1:02:38 - 1:04:45 FFTW: genfft finds good/new algorithms 1:04:45 - 1:06:30 FFTW: genfft's compilation strategy 1:06:30 - 1:08:00 FFTW: DAG creation 1:08:00 - 1:12:40 FFTW: OCAML Cooley-Tukey FFT 1:12:40 - 1:16:25 FFTW: Rader's algorithm for prime-size DFT 1:16:25 - 1:20:20 FFTW: The simplifier 1:20:20 - 1:24:40 FFTW: Conclusions and ongoing work Lecture on May 08, 2002 0:00:00 - 0:14:45 Start, course administration 0:14:45 - 0:16:30 Smart Matter: Frontiers in Computation 0:16:30 - 0:21:30 Three C's of Computer Science 0:21:30 - 0:23:45 Smart Matter Vision: A New way to Build ? and Systems 0:23:45 - 0:25:30 MEMS: Coupling to the Physical World 0:25:30 - 0:37:30 An "Active Surface" 0:37:30 - 0:43:15 Hierarchical Distributed Control 0:43:15 - 0:47:45 Smart Matter: Coupling 0:47:45 - 0:57:00 Collaborative Sensoring: Acoustic Tracking 0:57:00 - 0:59:00 Smart Matter: Machine Diagnostics 0:59:00 - 1:11:00 PolyBot: A Modular Reconfigurable Robot 1:11:00 - 1:14:45 Trends in Control: MEMS + Distributed Coordination 1:14:45 - 1:30:00 Final Thoughts Lecture on May 13, 2002 0:00:00 - 0:00:34 Projects: Start, student project presentations: 0:00:34 - 0:20:47 Projects: Oskar Bruening, Jack Holloway, Adnan Sulejmanpasic: Matlab*P visualization 0:20:47 - 0:33:00 Projects: Ken Takusagawa: Tabulating values of the Zeta function 0:33:00 - 0:45:30 Projects: Nathan Warshauer: Parallel real-time Strategy AI testing 0:45:30 - 0:59:30 Projects: Ian Chan: Parallel 2D Kolmogorov-Smirnov statistic 0:59:30 - 1:16:00 Projects: Matt Craighead: Real-time parallel radiosity 1:16:00 - 1:24:00 Projects: Andrew Wilson, Ashley Predith: Simulation of oxygen ion ordering as a result of temperature change Lecture on May 15, 2002 0:00:00 - 0:01:15 Projects: Start, student project presentations 0:01:15 - 0:20:45 Projects: Ahmed Ismail, Cynthia Lo: Parallel off-lattice Monte Carlo simulations 0:20:45 - 0:32:45 Projects: Andrew Menard: Parallel clock designer 0:32:45 - 0:53:00 Projects: Amay Champaneria: Parallelizing condensation for visual tracking 0:53:00 - 1:12:00 Projects: Per-Olof Persson, Sreenivasa Voleti: Solving very large finite element problems in parallel 1:12:00 - 1:22:30 Projects: Dean Christakos: Linking Beowulf clusters across the grid