Differenze

Queste sono le differenze tra la revisione selezionata e la versione attuale della pagina.

--- magistraleinformaticanetworking:spm:detailedprogram0910 [20/01/2010 alle 09:40 (15 anni fa)] – creata Marco Danelutto
+++ magistraleinformaticanetworking:spm:detailedprogram0910 [20/01/2010 alle 12:43 (15 anni fa)] (versione attuale) – Marco Danelutto
@@ Linea 5: / Linea 5: @@
   * Evolution of computer architectures (multi/many core, clusters, grids, clouds) and urgency of effective parallel/distributed programming models and tools
   * Problems in classical parallel/distributed programming
+    * Distinction between distributed and parallel based on computational grain
+    * Definition of mapping and scheduling problems
+    * Communications and synchronization
+    * Computation grain
+    * Load balancing, reliability, security and autonomic management
   * Definition of structured parallel/distributed programming model
+    * Patterns/Skeletons
+    * Key features of a pattern
+    * User programmers and system programmer views
+    * Layered implementation
+    * Pros and cons
   * Skeleton/patterns for parallel/distributed programming
-    * techniques for discovering patterns
+    * Techniques for discovering patterns
-    * classification of patterns
+    * Classification of patterns (data, stream, control parallel)
-    * typical patterns
+    * Typical patterns
-  * Skeleton frameworks (definition of algorithmic skeleton)
+    * Cole's manifesto principles (with extensions)
+  * Skeleton frameworks
+    * Definition of algorithmic skeleton
+    * Language based vs. library based skeleton frameworks
   * Implementation of skeleton frameworks
-    * implementation templates
+    * Implementation template technology
-    * macro data flow
+    * Macro data flow technology
   * Fixed vs. variable skeleton set frameworks: how to handle expandability of the skeleton set using macro data flow implementation
   * Advanced features
-    * portability (virtual machine based vs. compiler based)
+    * Portability (virtual machine based vs. compiler based)
-    * adaptivity (behavioural skeletons with autonomic management of non functional features, hierachical, multiconcern)
+    * Adaptivity
+      * General principles
+      * Behavioural skeletons with autonomic management of non functional features
+      * Hierarchical management
+      * Multiconcern management
   * Performance modelling of structured parallel/distributed computations
-    * measures of interest (Service and completion time, speedup, scalability, efficiency)
+    * Measures of interest
-    * performance models for farm, pipeline and data parallel skeletons
+      * Service and completion time
-    * performance models using queue theory results
+      * Speedup and scalability
-    * how to derive a performance model for a generic skeleton.
+      * Efficiency, isoefficiency
-  * Design patterns: concept of design pattern, parallel design patterns, sample parallel patterns, perspectives and differences with algorithmic skeletons
+    * Performance models for farm, pipeline and data parallel skeletons
-  * Software components: definition, usage, usage in a structured framework, sample SCA code, perspectives and differences with algorithmic skeletons and design patterns.
+    * Performance models using queue theory results
-  * Workflows: definition, usage, perspectives and differences with algorithmic skeletons and design patterns.
+    * How to derive a performance model for a generic skeleton.
+  * Design patterns
+    * Concept of design pattern
+    * Parallel design patterns
+    * Sample parallel patterns
+    * Perspectives and differences with algorithmic skeletons
+  * Software components
+    * Definition
+    * Component composition (use/provide ports, other kind of ports)
+    * Usage, usage in a structured framework
+    * Sample component frameworks (Java Beans, Fractal, SCA). Sample SCA code
+    * Perspectives and differences with algorithmic skeletons and design patterns.
+  * Workflows
+    * Definition
+    * Typical usage
+    * Scheduling and fault tolerance
+    * Perspectives and differences with algorithmic skeletons and design patterns.
   * Low level frameworks to support structured parallel/distributed programming
-    * POSIX-TCP/IP (targets clusters and network of workstations, with tools, sample code, command line utilities, implementation of collective operations with UDP)
+    * POSIX-TCP/IP (targets clusters and network of workstations
-    * OpenMP (targets multicores, with sample code and sample implementation of structured code)
+      * Client server sample implementation
-    * RPC/RMI (targeting clusters and networks of workstations, with concepts and sample code in C (SUN rpcgen), Java (RMI) and ProActive)
+      * Tools and command line utilities (ssh, scp, rsync)
+      * Implementation of collective operations with UDP
+      * Sample code in C/C++ and Java
+    * OpenMP (targets multicores)
+      * Directives and run time primitives
+      * Sample code
+      * Sample implementation of structured code
+    * RPC/RMI (targeting clusters and networks of workstations)
+      * Concepts
+      * Sample code in C (SUN rpcgen), Java (RMI) and ProActive)
     * Measuring and modelling POSIX/TCP platform
-  * Wireless networks: concept, typical protocols, usage in structured context
+      * Mechanisms used to measure time (wall clock, clock, per process)
-  * Peer-to-peer techniques: concept, centralized, distributed and hierarchical approaches, typical algorithms, usage in structured contexts
+      * Simple modelling of TCP communications (approximating communication costs at the steady state)
+  * Wireless networks
+    * Concept
+    * Infrastructure and ad hoc modes
+    * Channel access protocols
+    * 802.11
+    * Usage in structured context
+  * Peer-to-peer techniques
+    * Concept
+    * Centralized, distributed and hierarchical approaches
+    * Typical algorithms and applications
+    * Usage in structured contexts
   * Client server model vs. peer to peer: perspectives
   * Concepts and perspectives on parallel programming from the Berkeley paper "A View of the Parallel Computing Landscape"