Intel says that it is ready to 'take our second crack' at the discrete GPU market with Battlemage, its new Xe2 GPU architecture that will debut inside of Lunar Lake-based laptops later this year.
In the latest episode of Intel's in-house "Talking Tech" series, Tom Peterson (or TAP as we call him in the industry) talked about what we can expect from Intel's next-gen Xe2 GPU architecture. The new Xe2 "Battlemage" GPU will feature a 50% performance upgrade, a bunch of architectural change, and the lessons learned from Xe "Alchemist".
TAP talked about the basis of Xe2 for Intel graphics is Lunar Lake, which is an mobile processor coming later this year, but the Xe2 Tile will be used for Battlemage: the company's next-generation Arc series discrete graphics cards.
During the Talking Tech episode, TAP said: "so our engineers have been very, very busy improving our Arc GPU to where we are now ready to take our second crack with Battlemage". Fighting words from Intel moving towards Q4 2024, and the year ahead with 2025 and mutliple new generations of graphics cards launching throughout the year from NVIDIA and AMD.
- Read more: Intel's next-gen Arc Battlemage 'Xe2' GPUs rumored to launch in Q4 2024
- Read more: Intel's next-gen Arc Battlemage 'Xe2' discrete GPUs rumored to be on TSMC's 4nm process node
- Read more: Intel Arc Battlemage X2 and X3 GPUs spotted: up to 448 EUs on a 256-bit memory bus
- Read more: Intel's next-gen Xe2 'Battlemage' GPU: 50% perf uplift, Lunar Lake first, discrete GPUs later
- Read more: Intel's next-gen Battlemage Xe2-HPG GPU with up to 24 GPU cores, 12GB VRAM spotted
Tom "TAP" Peterson said: "It's a scalable architecture. So this Xe2 architecture is obviously the basis of our graphics for Lunar Lake, which is integrated, but it is also going into our Battlemage next-generation discrete graphics. So Battlemage is gonna be a larger, bigger configuration with faster memory subsystems & bigger caches, and all the rest of it, and maybe even different clocks and different power envelopes but the fundamental building block, that Xe2 core, is going to be the same".
TAP continued: "Xe core from generation to generation is gonna improve primarily on perf per watt and pref per area. And those two characteristics drive both our better performance for integrated when you're power-constrained and better performance for discrete when you're less power-constrained but more like how big a chip can I build. So our engineers have been very, very busy improving over our Arc GPU to where we are now ready to take our second crack with Battlemage".

On the memory side of things, the listed Arc Battlemage Xe2 GPU had 6 DRAM channels at 32-bit per channel = 192-bit memory bus in total. This GPU in particular had 12GB of VRAM (which should be GDDR6) at 19Gbps, with up to 456GB/sec of memory bandwidth.
This is up from the 17.5Gbps GDDR6 on Arc Alchemist, but it featured a 256-bit memory bus. Intel will have higher-end Arc Battlemage GPUs with higher-end 256-bit memory buses (and 16GB of GDDR6) offering up to 608GB/sec of memory bandwidth from the same 19Gbps GDDR6 memory modules, which should result in 9% more bandwidth.
Intel itself has promised a 50% performance improvement with Battlemage compared to Alchemist, but we'll get our first taste of Battlemage with Lunar Lake later this year before discrete Battlemage launches later this year and into more so into 2025.