Intel unveils oneAPI software stack for heterogeneous architectures

Chipmaker Intel has launched a software stack, oneAPI, which is a unified and scalable programming model designed for high performance computing and artificial intelligence based solutions.

Intel announced the stack at Supercomputing 2019 event held in Colorado, US. The company is also making additions to its data-centric silicon portfolio along with a new software initiative that represents a move away from single-architectures and single-vendor programming models. 

With the launch of oneAPI, Intel claims it is on an industry initiative to deliver a unified and simplified programming model for application development in heterogeneous processing architectures. The supported architectures include the likes of CPUs, GPUs, field-programmable gate array or FPGAs and similar accelerators. 

 “HPC and AI workloads demand diverse architectures, ranging from CPUs, general-purpose GPUs and FPGAs, to more specialized deep-learning NNPs,” said Raja Koduri, senior vice president, chief architect, and general manager of architecture, graphics and software at Intel. 

Koduri added that Intel was striving towards simplifying the customers’ ability to harness the power of diverse computing environments. 

oneAPI platform will provide developers the power to work on the architecture of their choice, while addressing the problems of separate code bases, multiple programming languages and different tools.  Intel also stated that oneAPI will preserve existing software investments while still providing flexibility to create versatile applications.

Inclusive of an Intel beta product, the specifications of the oneAPI includes a direct programming language along with APIs in a low-level hardware interface.  The product also provides developers with a portfolio of developer tools such as compilers, libraries and analyzers, which are packaged into domain focused toolkits. 

Along with oneAPI, Intel also unveiled a new category of general purpose Graphic Process Units based on Intel’s X architecture. 

Called the “Ponte Vecchio,” the GPU is built for high performance computing modelling, simulation workloads and AI training. Ponte Vecchio is built on Intel’s 7nm technology and will use Foveros 3D( 3D packaging integrated circuit) and EMIB(Embedded Multi-die Interconnect Bridge ) packaging innovations. 

The GPUs also feature extra technologies such as high-bandwidth memory, Compute Express Link interconnect ( a high speed CPU interconnect, and other intellectual properties.