The company has worked with microprocessor firm AMD to provide support for OpenCL in multicore design incorporating graphics procesors (GPUs) and CPUs.
“We have worked closely with AMD to produce the fastest and most developer-efficient way to delivering optimised code for heterogeneous multicore environments, with full support for OpenCL standards and portability across platforms,” said Curtis Davis, co-founder, v-p of engineering and chief operating officier at MulticoreWare.
“AMD firmly believes industry standards such as OpenCL allow the development community to fully leverage the benefits of heterogeneous computing platforms,” said Raghu Rao, senior director, developer solutions at AMD.
The intention is to allow developers to use these tools to take advantage of integrating a GPU with a CPU multicore systems for video and image analytics and gesture processing.
Global Memory for Accelerators (GMAC) enables development of code for CPUs and GPUs in a single pointer space, without requiring the developer to be aware of, or manage, the different memory domains and allocations.
There is a task manager for sequencing operations across both CPUs and GPUs, with resulting optimisation of memory accesses.
“By working with MCW to make key development tools freely available to developers worldwide, AMD is further advancing the role of heterogeneous computing and industry standards among the developer community,” said Rao.
The tools are available on Microsoft Windows platform initially, with support for Linux and MacOS coming in the subsequent weeks. AMD and MCW are both committed to global, open standards, with on-going participation and leadership through actions with Khronos.