SC19: Intel GPU Architecture With HPC and AI Acceleration, and oneAPI Software Stack

At SC19, Intel Corp. unveiled its vision for extending its leadership in the convergence of HPC and AI with additions to its data-centric silicon portfolio and a new software initiative that represents a paradigm shift from today’s single-architecture, single-vendor programming models.

Aurora system at Argonne National Laboratory will be the first US exascale system to leverage the breadth of Intel’s data-centric technology portfolio, building upon Xeon Scalable platform and using Xe architecture-based GPUs, as well as Optane DC persistent memory and connectivity technologies.
(Credit: Intel Corporation)

Click to enlarge

Addressing the increasing use of heterogeneous architectures in HPC, the company expanded on its existing technology portfolio to move, store and process data more effectively by announcing a new category of discrete general-purpose GPUs optimized for AI and HPC convergence. The firm also launched the oneAPI industry initiative to deliver an unified and simplified programming model for application development across heterogeneous processing architectures, including CPUs, GPUs, FPGAs and other accelerators. The launch of oneAPI represents millions of Intel engineering hours in software development and marks an evolution from limiting, proprietary programming approaches to an open standards-based model for cross-architecture developer engagement and innovation.

“HPC and AI workloads demand diverse architectures, ranging from CPUs, general-purpose GPUs and FPGAs, to more specialized deep-learning NNPs, which Intel demonstrated earlier this month,” said Raja Koduri, SVP, chief architect, and GM, architecture, graphics and software, Intel. “Simplifying our customers’ ability to harness the power of diverse computing environments is paramount, and Intel is committed to taking a software-first approach that delivers a unified and scalable abstraction for heterogeneous architectures.“

oneAPI: Developer-centric approach to heterogeneous computing
The oneAPI initiative will define programming for an increasingly AI-infused, multi-architecture world. It delivers an unified and open programming experience to developers on the architecture of their choice without compromising performance and eliminating the complexity of separate code bases, multiple-programming languages, and different tools and workflows. It preserves existing software investments with support for existing languages while delivering flexibility for developers to create versatile applications.

oneAPI includes an industry initiative based on open specs and an Intel beta product. The spec includes a direct programming language, APIs and a low-level hardware interface. The oneAPI beta software provides developers a comprehensive portfolio of developer tools that include compilers, libraries and analyzers, packaged into domain-focused toolkits. TIt release targets Xeon Scalable processors, Core processors with integrated graphics, and FPGAs, with additional hardware support to follow in future releases. Developers can download the oneAPI tools, test drive them in the firm’s oneAPI DevCloud
Learn more about oneAPI

Data-centric strategy delivers foundation for AI/HPC convergence
The company’s silicon portfolio is comprised of a diverse mix of architectures deployed in a range of silicon platforms. The foundation of Intel’s data centric strategy is the Xeon Scalable processor, which today powers over 90% of the world’s Top500 HPCs. IThe processors are the only x86 CPUs with built-in AI acceleration that are optimized to analyze the massive data sets in HPC workloads.

At SC19, the company unveiled a new category of general-purpose GPUs based on Intel’s Xe architecture. Code-named Ponte Vecchio, this high-performance, flexible discrete general-purpose GPU is architected for HPC modeling and simulation workloads and AI training. Ponte Vecchio will be manufactured on Intel’s 7nm technology and will be the firm’s first Xe-based GPU optimized for HPC and AI workloads. It will leverage Intel’s Foveros 3D and EMIB packaging innovations and feature multiple technologies in-package, including high-bandwidth memory, Compute Express Link interconnect and other IP.

Building foundation for Exascale computing
The company’s data-centric silicon portfolio and oneAPI initiative lays the foundation for the convergence of HPC and AI workloads at exascale within the Aurora system at Argonne National Laboratory. Aurora will be the first US exascale system to leverage the full breadth of Intel’s data-centric technology portfolio, building upon the Xeon Scalable platform and using Xe architecture-based GPUs, as well as Intel Optane DC persistent memory and connectivity technologies. The compute node architecture of Aurora will feature 2 10nm-based Xeon Scalable processors (code-named Sapphire Rapids) and 6 Ponte Vecchio GPUs. Aurora will support over 10PB and over 230PB of storage. Aurora will leverage the Cray Slingshot fabric to connect nodes across more than 200 racks.