Later this year Intel will release a research project code-named Ct, short for C for Throughput, that will automatically make standard C and C++ compilers work with many-core processors — those with 16 to hundreds of cores. Ct will be part of Intel's Parallel Studio software-development toolkit and will enable programmers to employ all of the cores in many-core processors, ensuring they are running at full efficiency.
Traditionally, creating parallel software has required determining which parts of the code were most easily adapted to parallel processors and isolating those parts in the module. However, isolating and applying parallel code would have to be done from scratch for each new processor family or with any significant increase in the number of cores. Intel says that Ct automates this process, and optimizing for many-core processors will not even require a recompile.
Ct also is backward compatible with software written for Intel's x86 architecture. Ct will work with Intel's new graphics chip, code-named Larrabee, due next year. "Ct is a good match for Larrabee," says University of Illinois professor Marc Snir. "We have thought of Ct as something that is much more attractive than CUDA or OpenCL for developing data-parallel code."
From Technology Review
View Full Article
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA