Intel moves toward modular GPUs with a new patent for chiplet-based architecture

Intel has filed a patent for a new GPU architecture that uses separate logic chiplets. The new disaggregated design is expected to improve scalability, computing power, and energy management.

In early October, Intel applied for a patent for disaggregated GPUs in a bid to continue its involvement in the tech and gaming spaces. The new design is Intel‘s latest attempt to shake up the GPU market currently dominated by AMD and NVIDIA.

The company is already one of the largest and oldest multinational technological companies in the US. It is known for its computer components, which include microprocessors, chipsets, and graphics processing units (GPUs). 

Intel’s patent for new GPU design

Earlier this month, Intel was finally granted a patent for its disaggregated GPU architecture, which will likely be the first commercial GPU architecture with logic chiplets, also allowing for the power-gate of chiplets not used to process workloads. pic.twitter.com/XsNjjdVIOu

— Underfox (@Underfox3) October 26, 2024

Intel was granted a patent for its new disaggregation of system-on-chip (SOC) architecture after it filed a request on October 8th, 2024. 

The patent was filed under Intel’s headquarters in Santa Clara, California. The new design is expected to be the first commercial GPU with logic chiplets. The new architecture is expected to introduce new features that will help improve the performance and power efficiency of the GPUs.

The patent describes a new method through which the system-on-a-chip (SoC) architecture can be broken down into separate components called “chiplets.” In this setup, every chiplet is an integrated circuit containing specific units of logic, partially packaged on its own. The chiplets, in this case, can be combined on a shared base. By using different chiplets with various core logic, manufacturers can assemble them into a single, larger device with varied functions.

The new design is stipulated to make the chips more flexible, allowing components to be mixed and matched for different tasks. The design also allows the GPUs to selectively activate or deactivate sections of the chip as needed.

However, the manufacturing complexity and costs to consumers are challenges that Intel will have to overcome before the chips are a profitable venture. Intel previously struggled with ramping up production of its Raptor Lake CPUs and user complaints about the costly nature of its GPU lineup. 

Intel hopes to disrupt the GPU landscape with new patent

The new patent filed represents Intel’s latest foray into the GPU market. Intel has lagged behind AMD and NVIDIA over the last few years. However, the new disaggregated GPUs have stirred interest among gamers and tech enthusiasts, who are eager to see how the new Arc Battlemage GPUs, set for an early 2025 release, will measure up.

AMD was also noted to have filed patents for disaggregated GPUs. The patents are stipulated to indicate the next competitive frontier in the GPU market.

Tech enthusiasts and gamers online speculated that modular designs may become the new industry trend, with manufacturers aiming to meet growing market demand for high-performance, power-efficient, and customizable GPUs.