News & Analysis
Comment
LarryM99
This is a start on making the power of multiple CPU cores accessible to ...
Intel upgrades parallel programming suite
Rick Merritt
9/2/2010 10:00 AM EDT
SAN JOSE, Calif. -- Intel Corp. has released a new version of its parallel development tool suite for desktop PCs. Parallel Studio 2011 incorporates new capabilities Intel acquired with startups Cilk Arts and RapidMind.
The suite targets Windows C and C++ code and is a plug in for Microsoft Visual Studio, aiming to speed development of parallel apps for systems using two to four x86 cores, the majority of the PCs Intel now ships. It represents one of many efforts to help code make use of multicore processors.
The Intel suite includes an expanded template library of C++ threading building blocks focused on task-level parallelism. It also includes a beta release of C++ array building blocks for data parallelism using both technology developed at Intel Research and at RapidMind. The release marks Intel's first use of technology from the startup acquired in August 2009.
In addition, the suite includes C language compiler extensions developed by Cilk Arts for serving both task and data parallelism. The latest version of the program also includes a so-called Advisor which can suggest ways to implement parallelism on a given program and check correctness of parallel code.
In the past, we only had threading building blocks and [support for] OpenMP," said Bill Savage, at Intel. "Now we feel we have a complete solution for parallel developers," he said.
Future releases will support managed code for languages such as Microsoft's .Net and C#. The tool is also expected to provide more support for heterogeneous parallelism in the future. Both Intel and AMD are rolling out processors next year that put x86 and graphics cores on the same die.




LarryM99
9/2/2010 11:53 AM EDT
This is a start on making the power of multiple CPU cores accessible to programmers, but it seems to me that we might have to move to a different programming model before it really takes off. Is anyone still working on the AI languages like Prolog and Eiffel anymore? How about F#? Do these hold promise for effective programming in multiprocessor systems?
Larry M.
Sign in to Reply