Closing out the last of the major PC-focused keynotes at Computex 2024 this evening, we have Intel. The long-reigning leader of the PC CPU market, Intel is in the middle of executing its plans to get back on track on both the manufacturing and chip design aspects of the business. Tonight’s keynote, being helmed by the highly-animated Pat Gelsinger, is titled “Bringing AI Everywhere.” And, like so many other Computex presentations and announcements this week, AI hardware is going to play a big part, as Intel outlines a full stack of products for client and server computing.
Of the four great PC chip vendors at the show, Intel has been the most up-front about what to expect from their hour-long presentation. The company’s Computex 2024 page already outlines their four major topics: AI PCs, Xeon 6 Processors, Gaudi AI accelerators, and Intel’s OpenVINO software ecosystem.
On the consumer hardware front, the company set the table with a significant teaser earlier this month about their forthcoming mobile PC SoC, Lunar Lake. The next generation of Core Ultra processors, Intel is touting significant energy efficiency gains for the mobile-focused chip, with new architectures driving their Performance and Efficiency CPU cores, Xe2 GPU, and a much faster 45+ TOPS (INT8) NPU. While the Lunar Lake announcement is coming relatively soon after the Meteor Lake launch, Intel has made it clear that it’s not going to hold back on shipping future products; they are looking to make up for lost time. Still, Lunar Lake devices are not expected to hit retail shelves until Q4 of this year, so this announcement is coming months in advance of the hardware itself.
On the server front, Intel has been publicly prepping for the launch of a new generation of Xeons with the Xeon 6 platform. The most notable part of this being the release of the company’s first Efficiency-core Xeon, Sierra Forest. Sierra Forest is set to be the first Xeon 6 chip out the door this year, and will offer up to 288 E cores on a single chip, allowing Intel to tap into the many (many) core CPU markets that AMD and Arm-based rivals have been unopposed at thus far.
Finally, the company has fully pivoted its server AI accelerator strategy to its Gaudi accelerators. Gaudi 3 was introduced back in April, and while it isn’t expected to go toe-to-toe with NVIDIA’s top accelerators in every workload, Intel is betting that they can beat NVIDIA on critical workloads, all while undercutting them significantly in pricing. The first Gaudi 3 parts are set to be released in the second half of this year, so hopefully we’ll be hearing a bit more about Intel’s plans as part of their keynote.
10:58PM EDT – Drumming, very loud drumming might I add
11:00PM EDT – Well that was, interesting to say the least. It’s gone dark..
11:00PM EDT – Welcome to the Intel Computex 2024 keynote, we are bringing you what Intel is doing live, with Pat Gelsinger expected to take the stage any minute. First, a video
11:01PM EDT – And here we go, Intel CEO, Pat Gelsinger, and the crowd are applauding massivelt
11:02PM EDT – Pat is talking about the Taiwan culture, but it didn’t take long for AI to get mentioned
11:03PM EDT – Touching on Moore’s Law, and talking about the first Computex back in 1981 (before I was born)
11:03PM EDT – The first trillion transistors on a chip, all the way back then. Looking to a trillion transistors by 2030
11:04PM EDT – Intel are having a 40th birthday party, but this year marks 39 years of Intel and Taiwan together. And a video to celebrate this partnership
11:05PM EDT – Unfortunately, the history is in the past, but the AI generation is now, and “every device will become an AI device, every company an AI company”
11:05PM EDT – Intel claiming they are 100% in the AI TAM
11:06PM EDT – Intel talking on their Foundry business, and driving the AI ecosystem. Enabling AI Everywhere is the theme of today’s keynote
11:07PM EDT – The AI PC, Edge and Enterprise, and the Data Center, all key focal points for Intel and the AI market share they want
11:07PM EDT – Pat is starting with the data center, enabling more performance in the cloud, over 130 million Xeons in the cloud currently
11:08PM EDT – Data center needs more compute, better power efficiency, but Pat is talking about issues such as the demands. Intel is launching the Solution
11:09PM EDT – Intel Xeon 6 (with E-Cores), Pat is weilding a Xeon 6 die, and Intel’s first Intel 3 based silicon
11:09PM EDT – Intel showing off Granite Rapids, and showing a visual demonstration
11:09PM EDT – Intel has 2 racks, first rack has 20 2nd Gen Xeon Scalable chips, the 2nd rack is empty
11:10PM EDT – Chuck is currently filling the second rack, with blades, to show the space saved by upgrading to Xeon 6
11:10PM EDT – And he’s done, five blades vs the 1st rack; a good visual representation
11:11PM EDT – The internet is 60-80% media
11:11PM EDT – A quick performance transcoding demonstration to showcase the performance of Intel Xeon 6 (Granite Rapids)
11:12PM EDT – Up to 4.2 X performance with Xeon 6 over the 2nd generation
11:12PM EDT – And that’s on E-cores
11:12PM EDT – 2. X performance per watts, 3 to 1 rack consolidation, and an 80k Megawatt power saving over time
11:12PM EDT – Over four years
11:13PM EDT – 144 Cores in a 2P solution with up to 288 cores, all E-cores, enabling a 6:1 consolidation ratio saving a lot fo rack space compared to 2nd Gen Xeon Scalable
11:14PM EDT – Talking about partners, Ebay, and rolling a video from SAP
11:14PM EDT – Primarily highlighting the power consumption savings with Xeon 6
11:15PM EDT – Moving onto the Ecosystem, partners including Supermicro, GIGABYTE, MSI, Microsoft
11:15PM EDT – Xeon 6 with P-Cores being shown off, it’s coming but not today
11:15PM EDT – And Pat Gelsinger is moving onto AI, in the enterprise sector and talking about LLM, challenges, and RAG
11:16PM EDT – 80% of the data remains on Prem
11:16PM EDT – Xeon + RAG changes AI workflows in the data center
11:16PM EDT – RAG stands for Retreival Augmented Generation
11:17PM EDT – Intel Gaudi 2, and the team are rolling on a system onto the stage
11:17PM EDT – Enterprise AI makes Gaudi 2 standardized for AI
11:18PM EDT – Intel running a Gaudi 2 demo with RAG + LLM, and it’s a Llama model for reference. Running a medical test (Xeon 6 + Gaudi 2)
11:18PM EDT – Running an AI analysis of a chest X-ray, using a multimodal LLM
11:19PM EDT – And running a Gaudi and Gaudi 2 price check
11:21PM EDT – Gaudi 2 AI kit costs 65K USD, with Gaudi 3 AI kit is much more expensive (125K USD)
11:21PM EDT – Gaudi 3 AI accelerator, 40% faster in time to train vs H100
11:21PM EDT – 2X the performance per $
11:21PM EDT – 15% faster Inference throughput than H100
11:22PM EDT – Pat showing off a wall of Gaudi 3 accelerators, from partners including ASUS, Lenovo, Inventec, Dell, GIGABYTE etc.
11:23PM EDT – Inventec joining Pat on stage, enter Inventec President Jack Tsai
11:24PM EDT – Jack Tsai talking about implementing Gaudi 3, and bringing it to the market, and will “fuel Gen AI”
11:24PM EDT – Gaudi is “highly scalable”
11:25PM EDT – Talking about PyTorch and how Gaudi 3 improves performance and power efficiency
11:26PM EDT – There’s growth in the AI service market, and Jack Tsai testements Gaudi 3 to fill this gap
11:26PM EDT – And back to Pat
11:26PM EDT – Now talking about interconnects, with network infrastructure for AI systems, with a focus on open tech and not closed solutions
11:27PM EDT – Touching on UALink, and Ultra Ethernet, much like AMD and Dr. Lisa Su did yesterday
11:27PM EDT – Which given it’s an open standard, having the two major players in Enterprise compute talk about it is big for adoption
11:28PM EDT – Intel say 50% of edge will run AI workloads
11:28PM EDT – We see “AI as a game changer” for the Edge
11:29PM EDT – Edge is the height of innovation, as they say, 50% of workloads on Edge will be AI by 2026
11:29PM EDT – Intel talking about putting AI into the hands of doctors, with Samsung Medical
11:30PM EDT – Intel talking about Core Ultra, SoCs for the mobile market
11:30PM EDT – “How many of you remember Wi-Fi?” Everyone Pat, everyone…
11:31PM EDT – Talking on Wi-Fi being adopted everywhere as a tool to access the internet. Pat says AI will be that Centrino level moment
11:32PM EDT – Intel shipped 8 million AI PCs so far. Estimated to ship 40 Million by end of 2024
11:32PM EDT – Pat unveils a wall of Intel Core Ultra notebooks and laptops. I’m blogging from Meteor Lake myself (Gavin)
11:32PM EDT – Pat showing off Lunar Lake, with a video roll from Satya. Same outfit, same background as every keynote so far
11:33PM EDT – Intel is unveiling Lunar Lake, 40+ TOPS NPU? That’s AMD Satya…
11:34PM EDT – Lunar Lake, Flagship SoC and Pat is showing the silicon off
11:34PM EDT – Lunar Lakes brings new IP blocks for CPU, GPU and NPU. Over 80 designs so far with over 20 OEMs
11:35PM EDT – Pat giving a side nod to TSMC, which Lunar Lake is built on (no Intel nodes this time)
11:35PM EDT – Lunar Lake has 50% better GPU performance, up to 4x AI perf, 40% lower SoC power, 3.5 X GPU AI Compute and up to 120 TOPS of overall performance (GPU, NPU, CPU)
11:36PM EDT – Intel putting discrete graphics on Lunar Lake
11:36PM EDT – Intel claiming their software support is more holistic than other chip makers in supporting software
11:37PM EDT – Pat talking about overall AI performance, NPU is important but not the most important. Up to 67 TOPS from the Xe2 graphics
11:37PM EDT – Lunar Lake outperforms XLE
11:37PM EDT – Sorry Qualcomm Snapdragon X Elite
11:37PM EDT – x86 can’t win on power efficiency, Lunar Lake busts this myth
11:38PM EDT – But Intel not disclosing ANY power figures. So take that with a pinch of salt, but we can be hopeful
11:38PM EDT – Lunar Lake isn’t launching today, however. Pat bringing ASUS to the stage, Chairman Jonney Shih
11:39PM EDT – To discuss new solutions, as ASUS and Intel are close partners
11:39PM EDT – Talking about AI PC, which is the theme of Computex 2024; Microsoft Copilot+ has driven us to this point today
11:40PM EDT – We’re expecting Jonney to start passionately shouting soon, much like he did during the AMD Keynote
11:40PM EDT – A Hybrid AI ecosystem, Jonney talking about the dynamics and how beneficial it is
11:41PM EDT – Unveiling the “Ultimate Weapon”, Lunar Lake. That’s a bold statement from ASUS
11:42PM EDT – ASUS Chairman “Jonney” is talking about their partnership with Intel over the years. Talking about harnessing the performance of Lunar Lake, Intel in the data center, through all their partnered avenues
11:43PM EDT – And he shouts!
11:43PM EDT – Intel talking about the client roadmap, Arrow Lake coming later this year, bringing AI to desktop in Q4 2024
11:44PM EDT – Panther Lake on Intel 18A, with Pat holding up a wafer, which is due to launch in 2025. The 5th Node in 4 years, and next week Intel claims they are powering on the first Intel 18A wafer
11:44PM EDT – Panther Lake on 18A, 2025
11:44PM EDT – Pat is welcoming Acer to the stage via Chairman and CEO Jason Chen
11:45PM EDT – Expecting this to be about Lunar Lake, as Jason Chen has an Acer notebook in hand, likely a Lunar Lake design
11:45PM EDT – Acer talking about shipping Meteor Lake, and claims they are getting amazing feedback
11:46PM EDT – “People are developing applications on NPU”; Well Lunar Lake brings way more performance, so…
11:47PM EDT – Customers are shifting from Search, to Ask
11:47PM EDT – “The AI computer is adapting to people”
11:47PM EDT – From people learning about computers, to computers learning about people
11:48PM EDT – Acer showing off a Lunar Lake based notebook
11:48PM EDT – Pat claiming that they will sell a load of Lunar Lake. The best is yet to come
11:50PM EDT – Pat talking about the new frontiers of innovation, sustainability, better outcoimes for everything in our lives
11:50PM EDT – The Intel Advantage, Global Scale with Open Standards
11:50PM EDT – “We believe in Moore’s Law”
11:51PM EDT – “Moore’s law isn’t dead, it’s alive in well” Said very emphatically by Pat Gelsinger
11:51PM EDT – Intel’s powerful ecosystem is a powerful one
11:51PM EDT – Rolling a video, everyone saying “I am with Intel”
11:52PM EDT – Pat saying thank you to all partners and the entire Taiwan Ecosystem. “Intel is with you, I’m with you”
11:52PM EDT – “Whatever has been done, will be outdone” Pat Gelsinger, 2024
11:53PM EDT – And that wraps up Intel’s Computex 2024 Keynote. Highlights include Gaudi, Xeon 6 with E-Cores, and Lunar Lake (Read our deep dive here – https://www.anandtech.com/show/21425/intel-lunar-lake-architecture-deep-dive-lion-cove-xe2-and-npu4)
11:54PM EDT – Thanks to everyone for joining us, see you next time!
