For those who like to build, it’s time for the second-ever Intel Innovation, a developer-focused technology showcase spotlighting the tools, training and community to empower the world’s developers to create what’s next.
Last year, Intel returned to its developer roots and pledged to empower an open ecosystem, ensure choice for developers, and build trust and partnership across the developer community. Intel CEO Pat Gelsinger opens the event this morning with a keynote to chart Intel’s progress on this pledge and to highlight new innovations and solutions designed to solve today’s challenges and open tomorrow’s opportunities.
8:45 a.m.: Hello and welcome! This is Jeremy Schultz, communications manager at Intel, and thank you for tuning in for another Intel Innovation. This year the event unfolds just down the street from Intel’s Santa Clara headquarters at the San Jose McEnery Convention Center, where I’m sitting with a laptop in hand and the big stage ahead.
8:55 a.m.: The hall is filling up! So nice to see so many faces fired up to have some high-tech fun.
9:01 a.m.: And in straight-forward developer style, we’re getting right to it: here’s Pat!
Wonderful to be in “Intel’s backyard,” Pat says, “to reaffirm Intel’s deep belief in an open ecosystem.”
9:02 a.m.: “We have a ton of news for you today,” he says. “We’ll geek out together with speeds, feeds and breakthroughs.” Bring on the toys! 🕹
9:04 a.m.: So let it be written: “We’re in a new era,” though “not quite post-pandemic,” Pat concedes.
“Technology is increasingly central to every aspect of human existence.” Even this simple liveblog requires quite a chain of infrastructure, come to think of it. 🤔
9:05 a.m.: We’ll enjoy “the true magic of technology” thanks to what Pat calls “the superpowers, foundational technologies…creating the bridge from the analog to the digital world.” Pat’s tremendous energy on stage is his superpower.
These superpowers “combine, amplify, and reinforce one another,” and they read in highway-sign-sized font across the stage: compute, connectivity, infrastructure, AI and sensing.
That last one is new to the club, Pat says, as technologies can see, identify objects, hear and even taste and smell. 👃
9:06 a.m.: The superpowers will also change how we think about chip design and manufacturing, Pat asserts, and the foundry model must evolve, too.
9:08 a.m.: What doesn’t change is that it’s all built on Moore’s Law. People have long doubted its longevity, but “Moore’s Law continues its relentless march forward,” Pat says. For Intel’s part, “we are ahead of our audacious schedule to deliver five nodes in four years.”
“Our 18A PDK 0.3” — translation: an early version of the “how to make a chip with this recipe” for the Intel 18A process node — “is now in hands of early design customers, and test chips are under design for end of year tapeout.” Tapeout means you send the design to be manufactured — you hit Control-P from a special computer in the factory (OK, it’s not really that simple).
9:09 a.m.: Still unclear? “Moore’s Law is alive and well,” Pat says.
And “Intel will be a great wafer foundry.” With Tower Semi, Intel Foundry Services “will be a globally diverse, end-to-end foundry with one of the broadest portfolios of technology.”
9:10 a.m.: “IFS will usher in the era of the systems foundry” — Cue the horns! — “marking a paradigm shift as the focus moves from system-on-a-chip to system in a package.”
This will be on the test: A systems foundry has four components, including the aforementioned wafers, packaging, software and an open chiplet ecosystem.
Novel packaging is giving designers “the tools to further increase the number of transistors per device.” Moore’s Law, turbo’d.
And software can accelerate product delivery, Pat suggests.
9:11 a.m.: Intel helped form the Universal Chiplet Interconnect Express (UCIe) consortium “to create an open chiplet ecosystem,” Pat explains. Imagine, for instance, if you got different chiplets from Intel, TSMC, TI and Samsung, and then had them assembled with packaging at Intel. You mix it, we match it.
Pat cues up a video with Dr. Kevin Zhang from TSMC and Jinman Han from Samsung, who each voice their enthusiastic thumbs-up for UCIe.
9:12 a.m.: “You just heard from the three largest chipmakers in the world, committing to a common direction,” Pat points out. “Amazing.” More than 80 companies in the semis industry have joined “to make an open chiplet interface a reality.” 🪁
9:13 a.m.: Funding can turbo, too. Intel’s $1 billion IFS Innovation Fund “directly invests in companies building disruptive technologies for the foundry ecosystem,” Pat explains.
“We’ve had a phenomenal response,” and among the first investments are Astera, Movellus and SiFive. By the numbers, in total: 150 companies engaged, 6 investments made, $400 million committed.
9:14 a.m.: And one more thing on foundry: Intel is expanding its university shuttle program with a goal to “scale our shuttles/passengers 100x.” 100x gets some applause! The idea is to “radically rebuild the semiconductor academic programs of tomorrow,” and give students and startups access to modern silicon technology. Whoa, that Control-P print-your-chip idea is almost true!
9:15 a.m.: It’s now time for “particular passion for me,” Pat says. “Graphics and accelerated computing.”
“We’ve known for decades” that the future of compute needs what GPUs provide, Pat says, “so I’m delighted we don’t have one GPU to show today, we have three!”
First up: the named-what-it-does Intel Data Center GPU Flex Series. “Pixels are growing exponentially,” Pat says. 🎥
9:16 a.m.: The Flex Series is a regular da Vinci of the visual cloud, excelling at media transcode, cloud gaming and AI. “We have an open and full stack approach” — are you sensing a theme? — “built on oneAPI.”
The news today, Pat says, is that “the Flex Series supports key industry AI and inference frameworks including OpenVINO, TensorFlow and PyTorch.”
9:17 a.m.: Next up: Ponte Vecchio, “our flagship high-performance GPU,” is now shipping to Argonne National Laboratory, along with 4th Gen Xeon processors, to form “the brain of the Aurora supercomputer.”
And the GPU story arc brings us to gaming and…Arc! Prices “in the sweet spot of desktop graphics” have nearly doubled in recent years, Pat says. “Gamers have been losing. Today, we’re fixing that.”
The A770 GPU, “our top-end gaming desktop GPU…delivers 65% better peak performance versus competition on ray tracing,” Pat explains. The A770 “will be available October 12…starting at $329.”
9:19 a.m.: Transition time! When Pat came to Intel at the age of 18, at “the temple of technology,” tech leaders “were maniacally focused about how technology would change the world.”
Pat invites to the stage “a young, bright technologist who embodies that passion to harness technology as a force for good,” and her name is Ria Cheruvu.
9:20 a.m.: At 18 herself, Ria is nearly eligible for her first Intel sabbatical, having started as an intern at 14 — the same year she received her undergrad from Harvard. 😳 You read that right.
Ria is an AI ethics lead architect, and works to make AI more usable, responsible and ethical.
9:21 a.m.: Intel’s got plenty of tools for that, Pat says, starting with the newly expanded Intel Developer Cloud. 🌩 This is not to compete with the big cloud vendors but rather “to win the hearts and minds of developers with early access to the latest Intel platforms for testing and evaluation.”
Starting in beta, pre-qualified customers and developers can fire up new and forthcoming Xeon processors, Habana AI accelerators and multiple Intel GPUs.
9:23 a.m.: Pat turns to Ria: you game to tackle some “pressing development roadblocks?” Ria: “I’m game, let’s do it!”
The first challenge is to speed up development cycles and break out of the “expensive and closed environments.”
9:24 a.m.: Demo expert Jon sets up the challenge: to speed up work of extracting code from projects hardcoded in CUDA.
The answer? “The Neural Coder plug-in that delivers AI performance optimization through a simple, one-click, no-code solution,” Jon says. He’s gonna show us live.
Jon and Ria take some computer vision code and give it a Neural Coder tune-up for 4th Gen Xeon chips, and performance goes from roughly 300 frames-per-second to well over 3,000 frames, over 10X better.
Just one click, Ria, you can retire now.
9:26 a.m.: Pat puts a new twist on this challenge: run this code “in a trusted execution context,” confidential computing style.
Jon returns with his toolbox and pulls out Gramine, which allows you to “create your own Intel Software Guard eXtensions (Intel SGX) enabled containers…with a few simple scripts.”
“Another one that’s just one click,” Ria says. And Pat calls it: challenge complete! 🏁
9:27 a.m.: Developer challenge #2: building AI models “without needing advanced degrees in AI modeling.” There’s only one Ria, after all.
Pat calls up another Intel colleague, Yannis Katramados, “inventor and product lead of a new, cool AI computer vision software platform.” 🥁
Yannis explains that AI model development is slow, expensive and few models make it to production. A lot of time and effort goes to waste, Ria agrees.
9:29 a.m.: Yannis and team have “a fresh approach: we have built an intuitive computer vision platform that enables enterprise teams to collaborate and build models very efficiently.”
Demo time! “Did you know coffee production is a $400 billion industry?” Wowza, I did not. Yannis says farmers could use computer vision to easily “count the ripe fruits and predict their yield.” ☕
9:31 a.m.: Building AI might typically take “tens of thousands of images,” Ria points out, but Yannis says a domain expert — somebody who knows coffee plants, in this case — might only need 10-20 images thanks to “active learning” technology.
Yannis annotates the fruit in the tool — immature, half-way-there, ripe. It’d take 5 minutes to train, so Yannis jumps to a trained model. He runs it, finds a couple errors, and corrects them. The final step is optimization with OpenVINO, which “means it can be deployed across different types of Intel hardware.”
9:32 a.m.: Ria: “This is so cool…all in this beautiful and easy-to-use interface.” 🎨
And Pat says it’ll be available soon. “Today we launch Intel Geti, powerful AI for everyone.” That’s geti.intel.com.
Geti lets “anyone on enterprise teams build computer vision models rapidly,” Pat says, and it’ll be “commercially available in Q4 for the enterprise.” Get Geti!
9:33 a.m.: Speaking of OpenVINO, “hundreds of thousands of developers are using it,” Pat notes, across a range of AI models and Intel hardware.
And Pat has a real-world example. He invites Hauke Feddersen from PreciTaste, a vision AI food management platform. Hallo Hauke!
9:34 a.m.: Hauke says Chipotle uses vision AI to guide crews what to cook and when based on demand and “freshness algorithms.” As a longtime Chipotle fan and sometimes frustrated manager of avocado freshness at home, I appreciate the complexity of this challenge. 🥑
The AI in the restaurant runs on a small PC like an Intel NUC, Hauke says. He applauds the “auto-optimization” OpenVINO provides and the “huge variety of devices to choose from” to run it.
9:37 a.m.: As Hauke shows how it works, Ria tries to trick the system by swapping out a different pan of food, but PreciTaste catches it.
Hauke says Chipotle is “piloting this computing one-stop-shop to scale,” and throws it to a video with Chipotle CTO Curt Garner. I’ll take a sofritas bowl with fajita veggies and a side of guac, thanks Curt! 🌯
9:38 a.m.: Pat thanks Hauke and Curt and Ria declares “this challenge is complete!”
Next up, Pat explains, is game development. Modern games are complex and take a long time to build. Pat invites Sarah Mainwaring from Inflexion Games to tell us more. 🎮
9:41 a.m.: Sarah and Inflexion are building a shared world survival crafting game called Nightingale. One of the challenges her team faces is fixing networking bugs in this kind of online multiplayer title. A tedious process when you can only run one instance of the game at a time.
But with a powerful enough system — like the one running on stage — Sarah can run as many as four copies of the game to better root out user-facing bugs. Backstage, Sarah says she was able to run eight!
9:44 a.m.: “As you just saw, the future of gaming and creating immersive experiences relies on performant CPUs and GPUs,” Pat says. Maybe Intel can help with that?
“I am thrilled to reveal the 13th Gen Intel Core desktop processor family,” which bring “the world’s best gaming experience.”
It does so with “massive performance gains,” Pat asserts, and the i9-13900K is “the world’s fastest desktop processor at 5.8 GHz. This is incredible!”
9:46 a.m.: Additional goodies: up to 24 total P- and E-cores; state-of-the-art wireless, connectivity and memory; and unlocked K SKUs bring “the world’s best overclocking experience.” 🚀
And if 5.8 GHz isn’t top-fuel-dragster enough for you, early next year Intel will release a version that hits 6.0 GHz “out of the box, in limited volumes,” Pat explains. “An industry first and huge milestone for client computing!”
Across all PC product segments, expect more than 50 processors and nearly 500 13th Gen designs. “We will give you scale and choice,” Pat adds.
And with that, gaming challenge, check.
9:48 a.m.: To quest #4: “creative development is time-consuming and limited.” Pat suggests to Ria they design a movie poster for this keynote in real time.
“AI is unleashing a wide array of new opportunities for the creative process,” Pat says, and invites demo expert Jon to show “some amazing tech from Habana Labs using Gaudi2,” Habana’s latest deep learning accelerator.
Jon fires up the Latent Diffusion text-to-image model and says it can generate 16 whatever-you-type images in three-or-so seconds.
9:49 a.m.: Pat’s first request: “movie poster.” Ria embellishes to “superhero movie poster,” and the system again awards an array of instant art. Capes or no capes?
Finally our duo asks for “superhero movie poster with male and female in art deco style.” Ria marvels at the speed. It’s a whole new “speed to creativity” for designers, Jon says.
And there’s another challenge knocked down.
9:50 a.m.: You know what Ria wants next? More real estate for different kinds of work, across the array of devices you might be carrying around.
Sounds like a challenge to me, Pat says, reaching into his top hat. I’m kidding, a top hat would be too small…unless the screen could roll up? 🗞
Pat invites JS Choi, CEO of Samsung Display, up to the stage. Welcome JS! (Excellent initials, my friend.)
9:53 a.m.: Pat saw this prototype on a trip to South Korea and “instantly saw the potential.” JS says it’s a great example of “the future of the PC.”
“Today we showcase the world’s first 17-inch slidable display prototype,” JS declares, and wow, it’s pretty wild…the screen can be simply stretched, like opening a window.
9:55 a.m.: As new form factors emerge, we’ll make it easy for software developers to make the best of them, Pat says. Speaking of software: “earlier this year at CES we announced acquisition of Screenovate, and today we’re excited to share we are bringing this product to market as Intel Unison.”
Demo expert Craig brings an Intel Evo laptop on stage and quickly pairs up and extends its display to the Android-powered roll-up screen.
Ria asks if it works with iOS and Craig quips that “Your Windows PC and iPhone will be singing and dancing together.”
9:57 a.m.: Intel Unison is “promoting an open ecosystem approach,” Pat explains, coming with new laptops this holiday season from the likes of Acer, HP and Lenovo.
9:58 a.m.: That’s now five developer challenges complete, leaving…the final challenge! “Bandwidth to silicon package is limited by physics.” Now we’re getting technical, folks. 🧮
Pat says we’ve been working on photonics, well, since before Ria was born.
9:59 a.m.: At Intel, “we invest for the long term,” and today, “our team has done it — our team built a detachable, optical-in-package connector.”
We’re going live to Intel’s lab in Scotland to see it for the first time ever!
10:00 a.m.: The team is bunny-suited-up and show a close-up of how the connector fits and how it plugs and unplugs. But does it work? Live charts show power and insertion loss, and if things go right, the power goes up and the insertion loss stays low.
The engineer connects the wire…and there was light! Applause erupts around the world. 🥳 Bandwidth curtailed by copper, overcome.
10:02 a.m.: Pat thanks Ria for taking the developer challenge and reminds her to pass on “our shared passion to change the world with Intel technology…in 30 years when you’re on stage as the CEO.” You got this Ria! 🏆
Ria is game for the challenge and promises to keep innovating. I don’t think this audience is going forget her poise on stage this morning.
“Let’s hear it for Ria!” 👏
10:03 a.m.: Pat’s got one more thing — a surprise guest instead of a surprise tech.
“Our collective potential is unlocked when we enable openness, choice and trust,” Pat notes, and this special guest is “someone I’ve admired for decades.”
There’s no bigger champion of open, collaborative systems than “the godfather of open source” and creator of Linux. It’s Linus Torvalds!
10:04 a.m.: Linus joins Pat on stage to cheers! Take us back to 1991, Linus, what was it like then?
I came from a home PC background and didn’t know x86, but I bought my first 386 and couldn’t afford the commercial Unix I wanted. I had this shiny new PC and was exploring the processor, I started on the beginning of a kernel.
10:07 a.m.: What’s a must-do for open source?
Linus: that people are invested in the community. I love open source and working in the community where it’s about getting things done together. I’m not a people person but at the same time, this community has really motivated me.
10:08 a.m.: Your vision for open source?
I’m not a visionary but I’m a plodding engineer. I leave that to others and focus on maybe the next 6 months of work.
10:09 a.m.: This might seem premature, Pat says to Linus, “but the Intel team agree that you are most worthy to be named the first recipient of the first-ever Intel Innovation Award for your lifetime of technical achievement.” 🏅
Linus: I don’t know what to say, this is so nice and classy. Thank you!
10:10 a.m.: As Linus trots off stage, Pat wraps things up. “Developers – whether software or hardware focused, you see the future. Our job at Intel is to open that world up to you.”
“Together we've peeked into the future, one we will create together. Thank you!”
And that’s a wrap for this live blog! Thank you for following along, and for all the details on Intel’s news today and through Intel Innovation this week, hop one link over to the Innovation press kit.
Notices and Disclaimers
Based on the Intel Core i9-13900K is the world’s fastest desktop processor at 5.8 GHz. As of September 7, 2022.
World’s Best Gaming Experience based on performance and unique features of 13th Gen Intel Core processors, including in comparison to 12th Gen Intel Core i9-12900K, AMD Ryzen 9 5950X, and AMD Ryzen 7 5800X3D, as of Sept. 7, 2022. See www.intel.com/PerformanceIndex for details.
Performance varies by use, configuration and other factors. Learn more on the Performance Index site.
For workloads and configurations visit www.Intel.com/PerformanceIndex. Click on the Events tab and Innovation Event Claims. Results may vary.
Overclocking may void warranty or affect system health. Results may vary. See www.intel.com/overclocking for details.
Intel® Unison™ solution is currently only available on eligible Intel® Evo™ designs on Windows-based PCs and only pairs with Android- or iOS-based phones; all devices must run a supported OS version. See intel.com/performance-evo for details, including set-up requirements. Results may vary.