Intel launches new HPC/Cluster development tools

Ahead of next week's supercomputing convention in Reno Nevada, Intel has launched a new set of tools to simplify cluster development for Linux and Windows optimized for Intel and AMD processors.  Key features are:

• Intel MKL adds an out-of-core solver, and includes cluster support (ScaLAPACK & DM-FFTs) standard in all copies
• Intel Cluster Toolkit has an edition which includes compilers (auto-installs on nodes)
• OpenMP support for mixing with Microsoft and gcc implementations of OpenMP
• Intel Trace Analyzer and Trace Collector extend support to OpenMP* on clusters
• Support in MPI analysis tools and the Intel MPI library for Windows CCS
• Support, in compilers and libraries, for Mac OS* X 10.5 (Leopard) and Xcode 3.0

I spoke with Intel's James Reinders (Director of Marketing for Intel Software Development Products) who explained that the new tools simplify cross platform MPI (Message Passing Interface) development.  Optimizations for Intel's new 45nm Penryn-class CPU have also been added to Intel's Compiler 10.1 and Reinders explained that they are trying to get their hands on an AMD Barcelona-based system but it's been a challenge getting Barcelona based system.

Advanced features like out-of-core memory management in the MKL (Math Kernel Library) allows programs to handle problems that are 10 to 100 times larger than physical memory in a graceful manner rather than relying on brute force virtual memory.  Reinders explained that the processors have gotten so fast that scientists are tackling mind boggling problems such as solving 25 million simultaneous equations which involves matrixes that are 25,000,000 by 25,000,000 in size.

Intel released the following pricing information: