Intel Tech Tour Taiwan 2024: Intel Introduces ‘Lunar Lake’ Platform for AI PCs
Intel plans to revolutionise artificial intelligence (AI) PCs with the debut of its new silicon.
The semiconductor giant at a media briefing ahead of Intel Tech Tour in Taipei, Taiwan introduced its new Lunar Lake client processors, which will power the next generation of AI laptops and bring new AI capabilities on a larger scale.
The new flagship processor boasts leaps in power efficiency, core performance, graphics and, of course, AI computing power over previous the previous generation. As with previous generations, these leaps all contribute to a PC experience that has productivity, creation, collaboration and entertainment functions at its foundation.
Firstly, Lunar Lake boasts up to 120 platform tera operations per second (TOPS) across CPU, GPU and NPU. The NPU, in particular, features over 48 TOPS, while GPU and the CPU feature up to 67 TOPS and five TOPS, respectively. These numbers enable the platform to facilitate Copilot+ experiences, which is the broad term used by Microsoft to describe PCs that have AI-accelerated capabilities and functions.
The upgraded NPU also features Up to six neural compute engines, up to 9MB of cache and 12 enhanced SHAVE DSPs.
Secondly, Lunar Lake’s CPU is designed to maximise performance and efficiency, incorporating an eight-core hybrid design that consists of four E-cores and four P-cores. The P-cores tout better single-threaded performance, 2.5MB of L2 cache per core and up to 12MB of shared L3 cache, while the E-cores tout 4MB of shared L2 cache and up to twice Vector and AI throughput.
What’s more, Lunar Lake has a new Xe2 integrated GPU, which features eight next-gen Xe cores, eight enhanced ray-tracing units and 8MB of cache, as well as XMX for AI throughput. The addition allows Lunar Lake to deliver up to 50% better graphics performance compared to Meteor Lake. It also powers new display and imaging engines and enables real-time raytracing, among a host of other tasks.
Lastly, Lunar’s Lake boasts some major improvements to energy efficiency and battery consumption. The new platform has up to 40% lower SOC power thanks to new PMIC power delivery, improved E-core cluster, memory side cache and enhanced Intel Thread Detector.
For connectivity, the platform has integrated Wi-Fi 7.0, Bluetooth 5.4, Thunderbolt 4 and PCIe Gen 5.0 and Gen 4.0. Meanwhile, for memory, it has up to 32GB with two ranks, up to 8.5GT/s per chip, support for LPDDR5X DRAM, support for 16b x4 channels, as well as up to 40% lower PHY power and up to 250mm2 of area savings.
Intel said it will be deploying Lunar Lake in more than 80 PC models across over 20 different original equipment manufacturer partners. The chip maker emphasised that when it develops platforms, it starts with the experiences it can deliver.
With Lunar Lake, Intel said its roadmap is built on a number of principles. First, it wants to supercharge the hybrid core to balance compute and power. Second, it plans to double down on Intel Arc, the chip maker’s brand of GPUs, to become a leader in vector and matrix computing. Third, it wants to optimise multi-engine AI and get the best results from every model and application. Fourth, it wants to lead advanced packaging. Finally, it looks to foster a vibrant ecosystem, where developers are welcome to create solutions, systems, software and tools on the chip maker's platforms.
Intel has unveiled the new Lunar Lake platform, which will power the next generation of AI laptops.
Lunar Lake boasts up to 120 platform tera operations per second (TOPS) across CPU, GPU and NPU, enabling the platform to facilitate Copilot+ experiences.
The platform also boasts leaps in power efficiency, core performance and graphics over the previous generation.