Intel is aiming to take on the likes of AMD and Nvidia with its new generation of consumer graphics cards, codenamed ‘Battlemage’.
Intel is primarily known for its work with CPUs, however in the past couple of years the tech giant has been dipping its toes into the GPU market. Sadly, its initial efforts with the first generation of Intel Arc graphics cards failed to match the performance or sales figures of AMD or Nvidia.
Intel is not out of the game yet. The company has its hopes pinned on its next-generation Arc GPU, codenamed ‘Battlemage’ which promises some considerable upgrades compared to its predecessor.
Intel is yet to confirm the line-up of Battlemage GPUs, so much of this information is derived from leaks and reports from industry insiders. As such, it should be taken with a pinch of salt.
Intel Arc ‘Battlemage’ release window speculation
At Embedded World 2024 in April, representatives and industry insiders on the show floor were heard to say that Intel plans for a late 2024 release window. The target date was indicated to be Black Friday, which this year will take place on November 29.
If accurate, this would put the Battlemage launch roughly two years after the release of the predecessor, the Arc Alchemist series.
The BMG-G10 card is expected to be the first to launch, with the BMG-G21 and BMG-G31 following later. However, some unsubstantiated rumors have suggested that the BMG-G10 has been canceled and the BMG-G21 and BMG-G31 – and thus the Battlemage line as a whole – have been pushed back to 2025.
Intel Arc ‘Battlemage’ leaked specifications
According to a leaked Intel roadmap, there are three planned cards. The BMG-G10, BMG-G21 and the BMG-G31.The most substantial specifications we have so far are for the BMG-G10, which was initially reported as the high-end flagship chip for the Battlemage line.The available specs are as follows:
- Xe2 HPG Architecture
- 56 Xe cores
- 448 XVE (EU)
- Each XVE unit comprises 2 shaders
- 112MB Adamantine cache
- 16GB 256-bit GDDR6X video memory
- Power consumption of up to 225W
Data acquired from the SiSoftware Sandra website by WCCFTech shows that Intel is working on two Xe2 chips for the Battlemage GPU. One of which features 20 Xe-Cores, while the other has 24 Xe-Cores. Both are listed as having 15GB of VRAM.
In June 2024, Videocardz reported that the BMG-G10 was set to have 28 Xe-Cores, contrary to the previously leaked figures. Reliable leaker Bionic_Squash on Twitter said that the BMG-G31 will be the largest of the Battlemage line, with 32 Xe2 cores.
Reported by Twitter user and industry insider harukaze5719 that information about the BMG-G31 was briefly posted on the Intel website. It was listed in a document marked “PTT Engagement – BGA3283 – BMG-G31 VRTT Interposer – Prototypes (Internal only).”
Subscribe to our newsletter for the latest updates on Esports, Gaming and more.
This document has since been removed.
Intel Arc ‘Battlemage’ performance speculation
On paper, the specs do not appear to be more impressive than the already existing Alchemist. However, the Battlemage is said to have a doubled core count per Xe-Core. In addition, Intel will be implementing a new memory subsystem. This could introduce GDDR7 support, which has great potential for improving performance.
It is also worth remembering that firmware and drivers can make a substantial difference in GPU performance. At launch, the Intel Arc alchemist did not produce impressive figures, but some users have since adopted the Arc platform thanks to improvements made by subsequent Firmware and driver updates.
According to leaker Bionic_Squash and reported by NotebookCheck, Intel intends to focus on the low-to-mid range of the GPU market. Since Nvidia discontinued the GTX 1630 and 1650, this has left a gap in the budget range that Intel may be aiming to take advantage of.
Intel Arc ‘Battlemage’ features speculation
When it comes to Nvidia vs AMD for gaming performance, the pendulum has swung towards Nvidia. This is due to its suite of features, namely DLSS. Intel, however, has its own answer to DLSS.
Thanks to its additional AI-powered grunt, Nvidia’s DLSS can beat out the performance figures posted by its AMD equivalents by taking advantage of features such as frame generation. This technology inserts additional frames during play to ensure a smooth gaming experience.
Intel’s version, named XeSS, uses a similar technique to Nvidia called Frame Extrapolation. Frame Interpolation, like that used by Nvidia and AMD’s systems, typically uses two frames of data, those that come before and after the inserted frame in order to mix together the added frames.
Frame Extrapolation, by contrast, only needs one frame of data. This means the system has to work harder to fill in the gaps, but it has one crucial advantage – it reduces input lag.
A version of this technology is planned for current Intel Arc GPUs, so upgrades and improved versions are likely planned for Intel’s next generation of graphics cards.