Auto vectorization of OpenCL* code with the Intel® SDK for OpenCL* Applications

January 29, 2012 11:00 PM PST


The Intel® SDK for OpenCL* Applications features an implicit vectorization module which boosts the performance of applications. The implicit CPU vectorization module seamlessly compiles the user's OpenCL* kernels to fully utilize the full 8 wide floating point SIMD processing, boosting the performance of user code without user intervention. The implicit vectorization module uses state-of-the-art vectorization algorithms based on up-to-date compiler research to ensure that the code performs.

More details on the implicit vectorization module can be found in this blog: http://software.intel.com/en-us/blogs/2011/09/26/autovectorization-in-intel-opencl-sdk-15/

The Intel® SDK for OpenCL* Applications implicit vectorization module is based on the LLVM compiler toolkit. We contributed major changes to the open source community for better support of Intel® architecture: http://blog.llvm.org/2011/12/llvm-31-vector-changes.html

Watch the OpenCL implicit vectorization module in a talk at the latest LLVM developer conference.

OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by Khronos.

* Other names and brands may be claimed as the property of others.