CHIP DESIGNER AMD has submitted its OpenCL accelerated parallel processing (APP) software development kit (SDK) 2.7 for certification, claiming it is the first fully compliant implementation of OpenCL 1.2.
AMD has been pushing OpenCL very hard, hoping to attract developers to make use of its CPUs and GPUs that support the language. Now the firm has submitted its OpenCL APP SDK 2.7 for certification, claiming it is the first fully compliant OpenCL 1.2 implementation for both CPU and GPU.
With AMD providing support for OpenCL 1.2, the firm was keen to point out some new features of the language including improved image/buffer map operations and the ability to transfer buffers before they are needed, along with improved buffer handling and protection.
The firm has also updated its Gdebugger, APP Profiler, Kernel Analyzer and APP ML software tools.
Mark Ireton, a senior manager of Product Application Engineering at AMD said, "I am also excited that AMD now supports both the C++ wrapper AP, and the AMD extension to support the C++ kernel language enabling complete application development using C++ capabilities, removing the need for much of the OpenCL API boilerplate function calls in the host code while at the same time and improving type checking of kernel parameters."
AMD's Gdebugger now supports Linux, while its APP Kernel Analyzer 1.12 supports Catalyst drivers between 12.1 and 12.4. AMD has also updated code samples to help developers get a foothold on modelling problems to take advantage of OpenCL.
AMD said it expects APP SDK 2.7 to receive certification and has made it available for download. µ
Sign up for INQbot – a weekly roundup of the best from the INQ