Languages and Compilers for Parallel Computing

Itisourpleasuretopresentthepapersacceptedforthe22ndInternationalWo- shop on Languages and Compilers for Parallel Computing held during October 810 2009 in Newark Delaware, USA. Since 1986, LCPC has became a valuable venueforresearchersto reportonworkinthegeneralareaofparallelcomputing, high-performance computer architecture and compilers. LCPC 2009 continued this tradition and in particular extended the area of interest to new parallel computing accelerators such as the IBM Cell Processor and Graphic Processing Unit (GPU). This year we received 52 submissions from 15 countries. Each submission receivedatleastthreereviewsandmosthadfour.ThePCalsosoughtadditional externalreviewsforcontentiouspapers.ThePCheldanall-dayphoneconference on August 24 to discuss the papers. PC members who had a con?ict of interest were asked to leave the call temporarily when the corresponding papers were discussed. From the 52 submissions, the PC selected 25 full papers and 5 short paperstobeincludedintheworkshopproceeding,representinga58%acceptance rate. We were fortunate to have three keynote speeches, a panel discussion and a tutorial in this year's workshop. First, Thomas Sterling, Professor of Computer Science at Louisiana State University, gave a keynote talk titled HPC in Phase Change: Towards a New Parallel Execution Model. Sterling argued that a new multi-dimensional research thrust was required to realize the design goals with regard to power, complexity, clock rate and reliability in the new parallel c- puter systems.ParalleX,anexploratoryexecutionmodeldevelopedbySterling's group was introduced to guide the co-design of new architectures, programming methods and system software.

Klappentext

This book constitutes the thoroughly refereed post-workshop proceedings of the 22nd International Workshop on Languages and Compilers for Parallel Computing, LCPC 2009, held in Newark, DE, USA, in October 2009. The 25 revised full papers and 5 revised short papers presented were carefully reviewed and selected from 52 submissions. The papers address all aspects of languages, compiler techniques, run-time environments, and compiler-related performance evaluation for parallel and high-performance computing and extend the area of interest to new parallel computing accelerators such as IBM Cell processor and Graphic Processing Unit (GPU) presenting the latest work of leading researchers and practitioners in the field.

Inhalt
A Communication Framework for Fault-Tolerant Parallel Execution.- The STAPL pList.- Hardware Support for OpenMP Collective Operations.- Loop Transformation Recipes for Code Generation and Auto-Tuning.- MIMD Interpretation on a GPU.- TL-DAE: Thread-Level Decoupled Access/Execution for OpenMP on the Cyclops-64 Many-Core Processor.- Mapping Streaming Languages to General Purpose Processors through Vectorization.- A Balanced Approach to Application Performance Tuning.- Automatically Tuning Parallel and Parallelized Programs.- DFT Performance Prediction in FFTW.- Safe and Familiar Multi-core Programming by Means of a Hybrid Functional and Imperative Language.- Hierarchical Place Trees: A Portable Abstraction for Task Parallelism and Data Movement.- OSCAR API for Real-Time Low-Power Multicores and Its Performance on Multicores and SMP Servers.- Programming with Intervals.- Adaptive and Speculative Memory Consistency Support for Multi-core Architectures with On-Chip Local Memories.- Synchronization-Free Automatic Parallelization: Beyond Affine Iteration-Space Slicing.- Automatic Data Distribution for Improving Data Locality on the Cell BE Architecture.- Automatic Restructuring of Linked Data Structures.- Using the Meeting Graph Framework to Minimise Kernel Loop Unrolling for Scheduled Loops.- Efficient Tiled Loop Generation: D-Tiling.- Effective Source-to-Source Outlining to Support Whole Program Empirical Optimization.- Speculative Optimizations for Parallel Programs on Multicores.- Fastpath Speculative Parallelization.- PSnAP: Accurate Synthetic Address Streams through Memory Profiles.- Enforcing Textual Alignment of Collectives Using Dynamic Checks.- A Code Generation Approach for Auto-Vectorization in the Spade Compiler.- Portable Just-in-Time Specialization of DynamicallyTyped Scripting Languages.- Reducing Training Time in a One-Shot Machine Learning-Based Compiler.- Optimizing Local Memory Allocation and Assignment through a Decoupled Approach.- Unrolling Loops Containing Task Parallelism.