Intel unveiled Tuesday some new technology it hopes will help it regain some ground lost to its rivals in the chip-making industry.
The company announced at the Computex annual expo in Taiwan a new generation processors for datacenters, pricing for its AI accelerator kits and an architecture for an AI computer chip.
Intel boasted about its new Xeon 6 chip, which will provide more performance and better power efficiency than the previous generation of processors for high-density scale-out workloads.
Intel claims that Gaudi 2 and Gaudi 3 AI Accelerator Kits are capable of delivering high performance at a third of the cost of comparable platforms.
Intel stated that “the combination of Xeon Processors with Gaudi AI Accelerators offers a powerful way to make AI cheaper, faster and more accessible.”
Next-Gen AI PCs
On the PC front, the company unveiled the architecture underlying Lunar Lake, which it sees leading the next generation of AI PCs, with no-compromise application compatibility and stingy power consumption — up to 40% lower than previous chip generations.
Lunar Lake is Intel’s codename for the next-generation mobile processor.
Rob Enderle said that Intel’s competitors were equally focused and performing well. The race was far from over. Enderle GroupA firm that provides advisory services in Bend, Ore.
“The competitive outcome could depend on who stumbles the first. And none of these vendor are stumbling right now,” he said to TechNewsWorld.
Intel has a very broad strategy that is executed well, said University of Pennsylvania Professor of Engineering Benjamin Lee.
“Notably, all of these processor designs — Xeon, Gaudi, Ultra — have been developed and are expected to ship ahead of schedule,” he told TechNewsWorld.
Full Spectrum Reach
Intel CEO Pat Gelsinger maintained that Intel is one of the only companies in the world innovating across the full spectrum of the AI market — from semiconductor manufacturing to PC, network, edge, and data center systems.
He stated that “our latest Xeon Gaudi Core Ultra and Xeon Gaudi platforms, coupled with the power of the hardware and software eco-system, deliver flexible, secure, affordable, and sustainable solutions to our customers, allowing them to maximize their immense opportunities.”
Intel will achieve its “AI anywhere” vision with the full spectrum of reach. Alvin Nguyen is the Senior Analyst at Forrester. He told TechNewsWorld: “AI everywhere represents a smart bet that different sizes of generative AI will become the norm, versus Nvidia’s ultimate performance approach and being applied to larger models and data.”
AI software manufacturers are also interested in this kind of reach. It’s an advantage, because it is a single architecture, so you can theoretically have an app running from the PC to the datacenter,” said Jack E. Gold. J.Gold AssociatesNorthborough, Mass., has a IT consulting company.
“If you are moving from an Arm processor on a smartphone, to a Nvidia processor in a datacenter with an application, it’s going to be a lot of work because you have two different architectures,” said he.
Intel Strives to Regain market Share
Intel hopes that its sixth generation Xeon chip will help it regain some lost share on the data center market. According to Reuters Intel’s share in the x86 chip market has declined 5.6 percentage point over the last year, to 76.4%. AMD now holds 23.6%.
Penn’s Lee said that data center computing was Intel’s biggest market. “Continued improvements in performance and energy efficiency will be crucial,” he explained.
He said that the concern was that general-purpose CPU designs like the Xeon were becoming increasingly commoditized. Engineers in the entire industry know how define and refine designs.
He added that “performance and power advantages depend not only on design, but also on the transistors used in the construction of this design.” AMD and other competitors are designing data center processors with advanced transistors. This is increasing competition in the space.
Intel Xeon 6 Efficient Cores (E-cores), for companies upgrading their infrastructure, allows 3-to-1 consolidation at the rack level to save costs and space. (Credit: Intel)
Shane Rau has been a semiconductor expert with IDC, a market research firm, reported that the new Xeon chip comes in two flavors – the 6700 series and 6900, which offer different combinations of performance, power consumption, and cost. This product segmentation recognizes the fact that end-user workloads have become more diverse. Depending on which workloads an end-user is dealing with, they might need to prioritize performance, or balance performance and energy consumption, he said.
He continued: “So, Intel is expanding the types of workloads that it can support and trying to improve their competitive position against other vendors of data center processors, including AMD’s CPUs and GPUs, and Nvidia’s server GPUs.”
He said that Intel was trying to position their server microprocessors so they could take on AI-based workloads themselves, and that users wouldn’t have to purchase a separate server graphics card to accelerate these workloads.
Lunar Lake: High Hopes
Intel has also high hopes for Lunar Lake. Mark N. Vena (President and Principal Analyst at): “Intel believes that Lunar Lake could catapult it back into a leading position against Qualcomm or AMD.” SmartTech Research TechNewsWorld reports that Las Vegas is a popular destination for tourists.
Lunar Lake chips are optimized for AI workloads and feature advanced AI integration on the hardware level. This sets them apart from AMD’s and Nvidia’s more traditional approach, according to him.
Vena says that the architecture is also a superior performer per watt. Focusing on energy efficiency it could outperform AMD Ryzen or Nvidia Grace in sustained workloads. Vena adds that a comprehensive ecosystem of software for AI development will be available, directly competing with Nvidia CUDA’s and AMD’s ROCm platform.
Lunar Lake is designed for PCs that are used by consumers, whereas Nvidia’s chips are more suited to data centers. AI applications will be easier to use and accessible. Intel’s Lunar Lake push could force AMD and Nvidia into integrating AI in their consumer products. This would drive competition and innovation within the AI PC market.
Gold predicts that 65 to 75 percent of PCs will have AI capabilities within the next two to three decades, and this number could even be higher in enterprises upgrading their machines. “So, what Intel does in that area is important,” said Gold.
He added, “But Lunar Lake doesn’t just concern AI.” It’s about reducing the power requirement because [consumers] Want thin, light machines that have batteries that can last for two, three, or even a full week.”
AI and Power Shortages: How to Address the Issue
Deborah Perry Piscione co-founder of The AI Institute and CEO, said that power is going to become a major issue for AI in the future. Work3 InstituteSan Francisco, California. She told TechNewsWorld that one of the most significant investments would be to upgrade our energy infrastructure, which is 40 years old.
“Policymakers need to act quickly in supporting AI development, and not be late like we were when it came to fabs for semi-conductors,” she added. “We paid the price when the Covid semiconductor shortage occurred, and we’ll continue to pay it on our energy network.”
“The success of AI development and our ability to consume it is all a big ‘if’ right now,” she said.
