Xe-LP

Xe-LP
Name	Xe-LP
Designer	Intel
Produced	From 2020
Predecessor	Gen11 graphics
Successor	Xe-HPG, Xe-HPC
Process	10 nm (Intel 7)
Products	Tiger Lake, DG1, Alder Lake

Xe-LP. It is a low-power microarchitecture for integrated and discrete graphics processing units designed by Intel and first introduced in 2020. The architecture marked Intel's first major step into the competitive discrete GPU market under the Xe branding, targeting laptops, entry-level graphics, and data center acceleration. Its design emphasized energy efficiency and a new execution unit structure to improve performance-per-watt over previous Intel Graphics Technology generations.

Overview

The Xe-LP microarchitecture was formally unveiled by Intel in 2020 as the foundational tier of its new Xe graphics roadmap. It was strategically positioned to serve the high-volume client computing market, powering the integrated graphics in Tiger Lake mobile processors and the company's first modern discrete GPUs like the Intel DG1. The development of Xe-LP was a direct response to the growing graphics demands in ultrabooks and the need for efficient media encoding and AI acceleration in thin devices. Its launch represented a significant architectural departure from the prior Gen11 graphics design, aiming to enhance Intel's competitiveness against offerings from AMD and NVIDIA in the mainstream segment.

Architecture

The Xe-LP architecture introduced a redesigned execution unit (EU) that increased the number of FP32 units per EU compared to its predecessor. These EUs were grouped into larger subslices and slices, forming a more scalable building block for both integrated and discrete configurations. Key architectural innovations included a new memory hierarchy with a larger L3 cache and support for GDDR6 memory on discrete products. The design also incorporated dedicated hardware for fixed-function tasks, such as an enhanced media engine supporting AV1 decode and DisplayPort 2.0 readiness. This modular approach allowed the same fundamental architecture to be implemented across a range of product lines from integrated SoCs to standalone graphics cards.

Product Implementations

Xe-LP first appeared as the integrated Intel Iris Xe graphics within Tiger Lake CPUs like the Core i7-1185G7, shipping in devices from Dell, HP, and Lenovo. Its first discrete implementation was the Intel DG1, a low-power graphics card for OEM systems and later retail add-in cards like the Iris Xe MAX. Xe-LP graphics were also integrated into later Alder Lake mobile processors and select data center products such as the Intel Server GPU, aimed at media transcoding workloads. These products were fabricated on Intel's 10 nm semiconductor node, later rebranded as Intel 7.

Performance and Features

In integrated form, Xe-LP graphics delivered significantly improved performance over Gen11 graphics, particularly in 1080p gaming and content creation applications, competing closely with AMD Radeon Vega mobile graphics. Its media engine was a standout feature, offering hardware-accelerated AV1 video decode, a first for the industry, which benefited platforms like Netflix and YouTube. The architecture also supported Intel Deep Learning Boost for AI inference tasks and featured advanced display capabilities for driving multiple 4K monitors. While its discrete DG1 offered modest gaming performance, it served as a crucial proof-of-concept for Intel's driver and software stack, including the oneAPI toolkits.

Manufacturing Process

Xe-LP products were manufactured using Intel's 10 nm SuperFin process technology, which later became known as Intel 7 as part of a node rebranding. This process provided significant improvements in transistor density and power efficiency over the previous 14 nm node used for Gen11 graphics. The use of an internal foundry process was a key part of Intel's strategy for its initial GPU offerings, though the company later announced plans to utilize external foundries like TSMC for subsequent Xe architectures. The 10 nm production was centered at Intel facilities such as those in Arizona and Israel.

Successors and Evolution

Xe-LP was succeeded by more powerful and specialized Xe architectures, including the gaming-focused Xe-HPG (used in the Intel Arc Alchemist series) and the high-performance computing Xe-HPC (used in Ponte Vecchio). These later architectures incorporated features like hardware ray tracing and AI upscaling via XeSS, moving upmarket to compete directly with NVIDIA GeForce and AMD Radeon RX series. The development of Xe-LP provided the essential foundation in driver development, software ecosystem, and manufacturing experience that enabled Intel's more ambitious entries into the discrete GPU market.

Category:Intel microarchitectures Category:Graphics microarchitectures Category:2020 in computing