Rumors are circulating about Intel’s upcoming Battlemage GPU potentially using TSMC’s 4nm process, as reported by DigiTimes and Ctee. This new graphics architecture, part of Intel’s Xe2-HPG lineup for desktop and mobile discrete GPUs, is expected to have 32 cores and improved ray tracing acceleration. The move to a smaller node could lead to increased complexity, higher clock speeds, and more advanced logic.
Intel’s current Arc Alchemist GPUs, manufactured by TSMC, feature up to 32 Xe cores. The rumored Battlemage GPU will also have 32 cores but with the updated Xe2 core. Additionally, Intel is reportedly considering using the 3nm node for their Celestial architecture (Xe3).
The Xe-HPG GPU (Alchemist) and Xe-LPG GPU (Meteor Lake) are already using TSMC’s N6 and N5 nodes, respectively. The upcoming Xe2-HPG GPU (Battlemage) is rumored to be utilizing the 4nm node, while the Xe2-LPG iGPU Processor (Lunar Lake) will be on TSMC’s N3 node. The Xe3-HPG GPU is expected to use the 3nm node.
While Intel has not provided any recent updates on Battlemage, they have introduced the Xe2 architecture as part of the upcoming Lunar Lake platform. Rumors speculate that Intel will release two variants of the Battlemage GPU: the BMG-21 with 20 cores and the BMG-31 with 32 Xe2 cores, expected to launch in 2025 at the earliest.
In 2025, other GPU vendors are also set to launch high-end and mid-range solutions, with NVIDIA aiming to introduce enthusiast GPUs in 2024. It seems the GPU market is gearing up for some significant advancements in the coming years with new architectures and smaller process nodes.
Article Source
https://www.fudzilla.com/news/graphics/59310-intel-s-flagship-gpu-will-be-4nm