Intel Is Ready To Take Its Second Crack At Discrete Gaming GPUs With Arc Battlemage “Xe2”: Larger Configs, Faster Memory & Bigger Caches

Intel is ready to enter the market with its 2nd Gen Arc discrete GPUs codenamed Battlemage, powered by the Xe2 core architecture.

Intel Has Learned A Lot From Its 1st Gen Arc Alchemist GPUs, Says Battlemage "Xe2" Built On The Foundation & Should Be Even Better With Bigger Configurations

In the latest episode of Intel's "Talking Tech" series, Thomas Petersen aka TAP discussed what to expect from the upcoming Xe2 GPU architecture, first featured on Lunar Lake SOCs & also coming to discrete GPUs in the form of Arc Battlemage. These GPUs are going to offer up to 50% uplifts in performance and feature loads of architectural changes while utilizing a solid software framework established with Alchemist chips.

Related Story Intel’s Next-Gen Diamond Rapids-AP “Xeon” CPUs Utilize Massive LGA 9324 Socket on “Oak Stream” Platform, 5 Times Larger Than LGA 1700

The work on Intel's Battlemage GPUs began as soon as the design team was finished with the Arc Alchemist architecture. The software-side team had to work on Alchemist for a bit longer as it was the first generation of discrete graphics from Intel. Before this, the company hadn't established a major ecosystem in the DIY segment since most of its architectures were integrated. It took some time but Alchemist's software support did ended getting the "Fine Wine" treatment and we still see good game support such as the recent drivers which added support for two high-profile titles that are launching this month.

With that said, all eyes are now on Arc Battlemage, bolstered by the Xe2 graphics architecture and offering increased performance and efficiency than the past generation. TAP states that Xe2-based Battlemage GPUs will be larger and bigger configurations than their integrated offerings on Lunar Lake SOCs.

These discrete GPUs will also feature faster memory subsystems and bigger caches & that's something to expect since discrete GPUs are less power constrained, leading to different clocks and power envelopes. We won't go