Intel made a variety of software announcements during Day 2 of the Intel Innovation event 2022 in San Jose, California, and the company reinforced its commitment to open software at a time when more chip companies are starting to focus more on the software tiers of the computing stack.
If Day 1 of Intel’s event was focused on GPUs and hardware, then Day 2 honed in on software, as the company launched new AI reference kits for healthcare, discussed new tools to help developers reduce time-to-market and increase performance and security, unveiled a quantum computing software development kit (SDK), and provided an update on progress of the open oneAPI specification.
Openness
Intel CTO Greg Lavender, who joined Intel in 2021 as CEO Pat Gelsinger was embarking on efforts to inject new life into the company, said during the event that Intel “is making good on its software-first strategy” with its commitment to open software ecosystems. He noted that about 90% of developers are using software developed or optimized by Intel, according to a Global Development Survey conducted by Evans Data Corp. in 2021.
Lavender also provided an update on the oneAPI standard, a cross-industry programming model for heterogeneous computing that allows developers to choose the best architecture for the specific problem they are trying to solve. He said Intel oneAPI 2023 toolkits will ship in December with support for Intel’s latest and upcoming new CPU, GPU and FPGA architectures, and will include tools like the open source SYCLomatic compatibility tool, which enables conversion of CUDA source code to SYCL source code, creating more flexibility for users. Lavender also said Intel subsidiary Codeplay will now assume responsibility for the oneAPI development community.
Intel also announced six more education and research institutions that have formed oneAPI Centers of Excellence to help grow support for the standard: the School of Software and Microelectronics of Peking University; the Science and Technology Facilities Council in the U.K.; Technion Israel Institute of Technology; the University of Utah in collaboration with Lawrence Livermore National Laboratory (LLNL); the University of California-San Diego; and the Zuse Institute Berlin.
AI
Regarding AI, Lavender and Red Hat CTO Chris Wright teamed up at the event to reveal that Intel and Red Hat are launching a joint AI developer program aimed at helping developers deploy AI models using Intel’s edge AI portfolio and Red Hat’s OpenShift Data Science, which is a cloud service for AI developers and data scientists.
Lavender also said that government contractor Leidos Health Group is building a proof-of-concept with Intel’s Project Amber software services for confidential computing that will be used to demonstrate protection of veterans’ health information for future use in mobile clinics. Liz Porter, president of Leidos Health Group, stated, “Project Amber liberates Leidos from the need to build and maintain complex, expensive attestation systems, allowing us to focus on our core differentiation like intelligent automation and AI/ML driven analytics.”
Quantum
On the quantum computing front, Intel joined a long list of companies that offer quantum computing SDKs. Anne Matsuura, director of quantum and molecular technologies at Intel Labs, said during a briefing before the Innovation event that the company’s quantum SDK is now available to some developers in a beta version that employs C++ programming language, a decision Intel made based on requests from university departments and other users.
“We expect to launch version 1.0 of the Intel quantum SDK in Q1 of next year,” she said. “That will include a simulation of the Intel quantum hardware and a target backend for simulations. Some of the applications that beta users have started using our SDK for are simulations in areas like fluid dynamics, which includes problems in aerodynamics, hydrodynamics, solar flares, linear systems of equations, and material simulation of chemical catalysts. For now, these are all small-scale simulations, but it is important that users are learning to program quantum algorithms using the SDK and learning to use Intel’s quantum technology.”
Making the quantum SDK available to a broader community of developers is another way the company is trying to promote open ecosystem efforts. “We need to encourage a larger group of people to become confident in programming quantum computers,” Matsuura said. “The Intel quantum SDK is kind of our entry into that race.”