Lunar Lake's iGPU: Debut of Intel's Xe2 Architecture

Intel has introduced its latest integrated graphics processing unit (iGPU) architecture, Xe2, with the launch of its Lunar Lake mobile chips. This new architecture is designed to enhance efficiency and performance in thin and light laptops, where integrated GPUs are crucial. The Xe2 architecture builds upon the previous Xe-LPG/HPG designs found in Meteor Lake chips and Intel's Arc discrete GPUs. The Lunar Lake iGPU features eight Xe Cores organized into two Render Slices, similar to its predecessor. Key improvements include a 50% increase in the hierarchical Z cache and a 33% increase in color cache capacity, aimed at optimizing performance for fixed function units. The Xe Cores have been restructured to feature longer vector widths, enhancing efficiency by reducing instruction control overhead. Despite these advancements, the compute throughput remains comparable to the previous generation, with Intel's iGPU still lagging behind AMD's Strix Point in terms of FP32 operations. However, Lunar Lake has made strides in FP64 throughput and reintroduced matrix units, which are beneficial for tasks like upscaling in gaming. The architecture also boasts a doubled L2 cache capacity, although it comes with increased latency. Overall, while Intel's Lunar Lake iGPU shows improvements in certain areas, it continues to face stiff competition from AMD's offerings.

- Intel's Lunar Lake introduces the new Xe2 iGPU architecture.

- The architecture aims to improve efficiency and performance for integrated graphics.

- Lunar Lake features enhancements in cache capacity and matrix units.

- Despite improvements, Intel's iGPU still trails behind AMD in compute throughput.

- The new design includes a doubled L2 cache but with increased latency.