Intel’s next generation of graphics cards is due to be launched at the end of this year, based on the Battlemage architecture that will be integrated into the next Lunar Lake SoCs due to arrive in the next few days. After a promising first generation but whose launch was a complete miss, Intel intends to strengthen its position in this market. In the last few hours, a leak has emerged concerning what could be the entry-level Battlemage card. The card is said to be equipped with 12GB of GDDR6 memory on a 192-bit bus, which translates into a bandwidth of 456 GB/s. It should be noted that Intel seems to be following the recommendations of several developers, including Epic Games, who believe that GPUs must now have at least 12GB of VRAM to take advantage of next-generation games.
The most eagerly awaited point is clearly the Xe2 architecture. Not only is better raw performance expected, but also improved energy efficiency, which could give Intel an edge over its competitors in terms of performance per watt.
Intel would position its Battlemage GPUs as direct competitors to mid-range to high-end graphics cards from AMD and NVIDIA. For the moment, we know nothing about the products that Nvidia will be offering, but in this segment, there will be nothing before 2025 from the greens. AMD, on the other hand, seems to be following exactly the same strategy as Intel: maximising the performance/price ratio for consumers who want to play in 1440p or 4K. Intel will therefore have to convince on the performance front, but also reassure users of the stability and maturity of its software ecosystem, an area where, despite a frantic race with multiple updates, the company has failed to make an entry…
A number of leaks and speculations have led us to project the following hypotheses about this Battlemage generation: A performance improvement of a factor of x1.5 to x2 compared with the previous generation. A major improvement in Ray Tracing acceleration capabilities , and the widespread use of faster memory throughout the range.
