OpenCL SDK 1.5 supports vectorization per CPU

OpenCL SDK 1.5 supports vectorization per CPU

Intel has released version 1.5 of the SDK OpenCL. The OpenCL implementation includes adjustments for Intel processors and will assist developers in programming applications parallelized for multi-core systems. With the release of the software now featured dominated the first with Intel's Sandy-Brige-generation processor Intel introduced SSE extension Advanced Vector Extensions (Intel AVX) - specifically to floating-point-intensive operations by up to a factor of accelerated eighth

This includes the OpenCL SDK generates 1.5 with Implicit CPU Vectorization a new module for the automatic vectorization of code , the self-matching SIMD instructions for Intel AVX. Developers should thus not about the intricacies of parallel hardware architecture look to benefit from the instruction sets supported by Sandy Bridge. The generated code is, according to Intel processor compatible with future generations.

Are also included in the SDK, the SDK OpenCL Offline compiler, a standalone tool with which you can build on Linux kernel offline (which is track the code-generation in its own window and can be analyzed). And for Windows developers have added Intel SDK for debugging plug-in for Visual Studio 2008 allow to set in which breakpoints in an instance of the kernel and read OpenCL various kernel and memory variables.

The Intel OpenCL SDK 1.5 is available for Linux systems (64-bit) and Windows (32 - and 64-bit) free to download. Intel Core and Xeon processors will also provide a pure runtime component.