Addressing the rising use of heterogeneous architectures in high-performance computing, Intel expanded on its current expertise portfolio to maneuver, retailer and course of information extra successfully by saying a brand new class of discrete general-purpose GPUs optimized for AI and HPC convergence. Intel additionally launched the oneAPI trade initiative to ship a unified and simplified programming mannequin for software growth throughout heterogenous processing architectures, together with CPUs, GPUs, FPGAs and different accelerators. The launch of oneAPI represents thousands and thousands of Intel engineering hours in software program growth and marks a game-changing evolution from at this time’s limiting, proprietary programming approaches to an open standards-based mannequin 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,” mentioned Raja Koduri, senior vp, chief architect, and basic supervisor of structure, graphics and software program at 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: A Developer-Centric Approach to Heterogeneous Computing
The oneAPI initiative Intel launched at this time will outline programming for an more and more AI-infused, multi-architecture world. oneAPI delivers a unified and open programming expertise to builders on the structure of their selection with out compromising efficiency and eliminating the complexity of separate code bases, multiple-programming languages, and totally different instruments and workflows. oneAPI preserves current software program investments with assist for current languages whereas delivering flexibility for builders to create versatile purposes.
oneAPI contains each an trade initiative primarily based on open specs and an Intel beta product. The oneAPI specification features a direct programming language, highly effective APIs and a low-level {hardware} interface. Intel’s oneAPI beta software program supplies builders a complete portfolio of developer instruments that embrace compilers, libraries and analyzers, packaged into domain-focused toolkits. The preliminary oneAPI beta launch targets Intel Xeon Scalable processors, Intel Core processors with built-in graphics, and Intel FPGAs, with extra {hardware} assist to observe in future releases. Developers can obtain the oneAPI instruments, take a look at drive them within the Intel oneAPI DevCloud, and be taught extra about oneAPI right here.
Intel’s Data-Centric Strategy Delivers the Foundation for AI/HPC Convergence
Intel’s silicon portfolio is comprised of a various mixture of architectures deployed in a spread of silicon platforms. The basis of Intel’s information centric technique is the Intel Xeon Scalable processor, which at this time powers over 90 % of the world’s High500 supercomputers. Intel Xeon Scalable processors are the one x86 CPUs with built-in AI acceleration which are optimized to investigate the large information units in HPC workloads.
At Supercomputing 2019, Intel unveiled a brand new class of general-purpose GPUs primarily based on Intel’s Xe structure. Code-named “Ponte Vecchio,” this new high-performance, extremely versatile discrete general-purpose GPU is architected for HPC modeling and simulation workloads and AI coaching. Ponte Vecchio can be manufactured on Intel’s 7 nm expertise and can be Intel’s first Xe-based GPU optimized for HPC and AI workloads. Ponte…
