Kathy Gibson reports – As technology assumes an even greater significant in everyone’s lives, Intel has reiterated its commitment to fostering and enabling open ecosystems.
“We in a new era that is not quite post-pandemic, but we can see a way to the other side,” says Intel CEO Pat Gelsinger in his Intel Innovation keynote last night. “And we can see how much of a role technology plays in our lives. In this digital era, we continue to witness the magic of innovation.”
He points out that these innovations are being driven by the foundational technologies of compute, connectivity, infrastructure with cloud and edge, and AI intelligence everywhere turning data into actionable insights.
“We think there is one superpower that we were missing before,” he adds. “This is sensing – the ability to see, hear, feel and taste things with augmented reality enhancing our human senses.”
While each of these “superpowers” is powerful on its own, when they work together, they reinforce, accelerate and unlock new and more powerful possibilities, Gelsinger adds.
“It changes how we think about our business and our role in the industry, he adds. “And, at Intel, Moore’s Law is at the heart of innovation. I have been talking about it for years, and Moore’s Law is not yet dead. In fact, we are ahead of schedule to get 1-trillion transistors in one package by the end of the decade.
“We will not rest until the period table is exhausted. We will continue to be the stewards of Moore’s Law into the future.”
Gelsinger presented a number of challenges that developers face — vendor lock-in, access to the latest hardware, productivity and time-to-market, and security, to name a few — and introduced a multitude of solutions to help overcome them, including:
* New and forthcoming technologies a click away in the Intel Developer Cloud: Starting in a limited beta trial, the Intel Developer Cloud has expanded to give developers and partners early and efficient access to Intel technologies — from a few months up to a full year ahead of product availability. During beta, selected customers and developers can try out and test many of Intel’s latest platforms in the coming weeks, including 4th Gen Intel(r) Xeon(r) Scalable processors (Sapphire Rapids), 4th Gen Intel(r) Xeon processors with high bandwidth memory (HBM), Intel Xeon D processors, Habana Gaudi2 deep learning accelerators, Intel Data Center GPU (code-named Ponte Vecchio) and Intel Data Center GPU Flex Series.
* Computer vision AI, built faster and easier: The new and collaborative Intel Geti computer vision platform (formerly Sonoma Creek) enables anyone in the enterprise — from data scientists to domain experts – to quickly and easily develop effective AI models. Through a single interface for data upload, annotation, model training and retraining, Intel Geti reduces the time, AI expertise and cost needed to develop models. With built-in optimisations for OpenVINO, teams can deploy high-quality computer vision AI within their enterprises to drive innovation, automation and productivity.
The Intel Developer Cloud, together with developer tools and resources designed to optimize performance including Intel(r) oneAPI toolkits and the Intel Geti platform, can help accelerate time to market for solutions built on Intel platforms.
Expanded technology portfolio
Gelsinger also used the occasion to present the latest advances across Intel’s product portfolio, including:
* A new standard for desktop processor performance: 13th Gen Intel Core desktop processors, led by the flagship Intel Core i9-13900K, help users better game, create and work with up to 15% better single-threaded performance and up to 41% better multi-threaded performance gen-over-gen (see separate article).
* A big step for Intel GPUs: Gelsinger provided updates across Intel’s graphics products, a key area of growth for Intel. Server blades with the Intel Data Center GPU, code-named Ponte Vecchio, are now shipping to Argonne National Laboratory to power the Aurora supercomputer.
* New workloads for Flex Series GPUs: The Intel Data Center GPU Flex Series, announced in August, gives customers a single GPU solution for a wide range of visual cloud workloads. It will now run popular industry AI and deep learning frameworks, including OpenVINO, TensorFlow and PyTorch. AI neuroscience customer Numenta, working in collaboration with Stanford University on real-world inference workloads on MRI data using Intel’s Flex Series GPU, is reporting amazing performance results.
* Intel Arc GPUs for gamers: Intel is committed to bringing back price and performance balance for gamers with the Intel Arc graphics family. The Intel Arc A770 GPU will be available in a variety of designs at retail starting at $329 on 12 October, providing compelling content creation and 1440p gaming performance.
* A high-fidelity AI boost for games: XeSS, or Xe Super Sampling, a gaming performance accelerator that works across Intel discrete and integrated graphics, is now rolling out to existing games through updates and will be available in more than 20 titles this year. The XeSS software developer kit is also now available on GitHub.
* Many devices, one experience: Intel Unison is a new software solution that provides seamless connectivity between phones (Android and iOS) and PCs — starting with functionality including file transfer, text messaging, phone calls and phone notifications — coming to new laptops starting later this year (see separate article).
* Data center acceleration, on demand: 4th Gen Intel Xeon Scalable processors include a number of accelerators for AI, analytics, networking, storage and other demanding workloads. Through the new Intel On Demand activation model, customers can turn on additional accelerators, beyond the base configuration of the original SKU, for greater flexibility and choice when needed.
Systems Foundry opens ‘new era for chipmaking
Leaders from Samsung and TSMC joined Gelsinger in his keynote to voice support for the Universal Chiplet Interconnect Express (UCIe) consortium, which aims to create an open ecosystem for enabling chiplets designed and manufactured on different process technologies by different vendors to work together when integrated with advanced packaging technologies. With the three largest chipmakers and more than 80 of the leading companies in the semiconductor industry joining UCIe, “we are now making it a reality,” Gelsinger says.
To lead this platform transformation enabling new customer and partner solutions with chiplets, Gelsinger explains that “Intel and Intel Foundry Services will usher in the era of the systems foundry,” with four major components: wafer manufacturing, packaging, software and an open chiplet ecosystem. “Innovation once thought impossible has opened entirely new possibilities for chipmaking.”
Intel previewed another such innovation in development: a breakthrough pluggable co-package photonics solution. Optical connections hold promise to enable new levels of chip-to-chip bandwidth, particularly in the data center, but manufacturing difficulties make them untenably expensive. To overcome this, Intel researchers devised a robust, high-yielding, glass-based solution with a pluggable connector that simplifies manufacturing and lowers costs, opening possibilities for new system and chip package architectures in the future.
Building the future requires software, tools and products, and it also requires funding. Early this year, Intel launched the $1-billion IFS Innovation Fund to support early-stage startups and established companies building disruptive technologies for the foundry ecosystem. Now, the company has announced the first round of companies that have received funding, a diverse group innovating across the entire semiconductor stack.
The first round includes:
* Astera, a leader in purpose-built data and memory connectivity solutions to remove performance bottlenecks throughout the data center.
* Movellus, which helps improve system-on-chip performance and power consumption, and simplifies timing closure with a unique platform that solves clock distribution challenges.
* SiFive, developing high-performance cores based on the open-source RISC-V instruction set architecture.