Image

Intel Next-Gen Lunar Lake CPUs to Power Local Copilot AI with 45 TOPS NPU Performance

6/13/2024

Intel Next-Gen Lunar Lake CPUs to Power Local Copilot AI with 45 TOPS NPU Performance

Intel recently unveiled its next-generation Lunar Lake processors at the Vision 2024 event in Phoenix, Arizona. These chips promise significant advancements in AI performance, with a total package performance exceeding 100 TOPS (Tera Operations per Second). Notably, 45 TOPS are dedicated to the Neural Processing Unit (NPU), a specialized component optimized for AI tasks.

What You Need to Know

  • Lunar Lake Performance: The upcoming Lunar Lake CPUs will offer more than 100 TOPS of total package performance for AI workloads.
  • NPU Focus: The NPU within Lunar Lake chips will deliver 45 TOPS, emphasizing Intel’s commitment to AI capabilities.
  • Comparison: Intel’s current Meteor Lake lineup, including Core Ultra CPUs, provides 34 TOPS in total performance, with only 10 TOPS allocated to the NPU.
  • Microsoft’s Copilot: Microsoft plans to enable its AI assistant, Copilot, to run locally on laptops. To achieve this, a minimum of 40 TOPS of NPU performance is required.

AI PCs on the Rise

The industry is witnessing a surge in AI-focused PCs, and Intel, AMD, and Qualcomm are all vying for supremacy in this space. TOPS serves as a key metric to gauge AI performance, reflecting the ability to handle complex computations.

  • Qualcomm’s Snapdragon X Elite: Qualcomm’s Snapdragon X Elite CPU boasts 45 TOPS of NPU performance and 75 TOPS across the entire chip.
  • AMD’s Efforts: AMD’s Ryzen 8040-series mobile chips offer 16 TOPS for the NPU, with a total package performance of 39 TOPS. Their next-gen “Strix Point” mobile CPUs are expected to triple AI performance using the new XDNA 2 NPU.
  • Intel’s Vision: Intel, AMD, and Qualcomm recognize the importance of AI performance, and the race is on to deliver powerful NPU capabilities.

Copilot Goes Local

Microsoft’s Copilot, an AI assistant, is set to run locally on AI PCs. Intel’s late-March AI summit revealed this development, promising benefits in security, reliability, accessibility, and cost reduction. Tom’s Hardware reports that achieving Copilot’s local execution requires at least 40 TOPS of NPU performance—currently matched by Qualcomm’s unreleased Snapdragon X Elite chip.

As we move forward, expect more options for AI enthusiasts as Intel and AMD continue to enhance NPU performance.