Several years after Intel announced it would return to the graphics arena, the company has launched its first discrete GPU — the Intel Xe Max. It’s not unusual for low-end parts and integrated GPUs to share common building blocks, but the discrete GPU packaged inside a Xe Max card is an Intel Tiger Lake GPU in nearly every respect. Same 96 EUs, 768 cores, 48 TMUs, and 24 ROPs. The only changes are the use of a dedicated 4GB VRAM buffer for the Xe Max, and that GPUs slightly higher clock speed (1.65GHz versus 1.35GHz on TGL).
Now, in theory, this could be great news for fans of multi-GPU gaming, but we have to dash your hopes — Intel has no plan to bring that kind of workload-splitting technology to market with this line of cards. Instead, Intel is positioning the Xe Max as a content accelerator, with the ability to speed up certain kinds of AI and GPU workloads while the rest of your system components are devoted to other tasks.
Intel, in other words, isn’t really positioning the Xe Max as a gaming GPU at all. The iGP should be somewhat faster than the built-in Tiger Lake solution because it has a 1.65GHz maximum clock compared with the 1.35GHz Tiger Lake parts. But the Xe Max is still limited to a 25W TDP. One rather unusual thing about the Xe Max is its use of LPDDR4X as memory (4GB total). This is the first time we’ve seen a laptop GPU equipped with that kind of VRAM, at least that I can recall.
Introducing Additive AI
But while Intel may not be planning to allow for dual-graphics gaming, it does have plans to leverage both GPUs in certain workloads. The company is working with vendors to support what it calls Intel “Deep Link” technology, which allows both GPUs to be used in the same application.
Intel is also extending its power-tuning Adaptix technology to support Xe as well, meaning the company can balance the TDP demands of both CPU, iGPU, and dGPU as one contiguous “unit.”
At least some of Intel’s overall power efficiency improvement is the result of better overall resource management.
There’s not a lot to say about the Intel Xe Max in terms of its gaming, so I’m not going to belabor the point. You don’t buy this card for a gaming GPU, but if you do want to game on it, it can broadly hang with Nvidia’s bottom-end MX350. Even so, I’d like to point out that it’s quite possible to play a lot of great games on modern solutions — it just means playing games that might be a few years old, or at 720p. Intel and Nvidia aren’t competing for gamers with these chips, but it’s possible to squeeze some gaming out of them, if you pick your titles right.
But Intel’s idea of using a secondary GPU as an AI processing unit isn’t crazy. I use the Topaz suite of applications on a near-daily basis, and they’d benefit from this kind of support.
There are only three laptops launching with the Xe Max, and Intel obviously isn’t putting some major gaming push on the topic, but this is still an important moment for Intel. The company has re-entered the GPU market, even if it’s only dipping a toe in thus far.
8K Displays Could Be Ready This Year, but Content Could Take Until 2025
8K panels could arrive as soon as this year (at extraordinary prices), but 8K content? That might be seven years away — or even more. This depends, to some extent, and how you define "content."
Facebook to Build Its Own Silicon for Real-Time Content Filtering
Facebook is reportedly going to start building its own hardware for content monitoring and real-time filtering.
Star Citizen Unveils $27,000 Content Pack, Because of Course
Star Citizen's financial campaigns are back in the news, this time for a $27,000 pack that costs a bit less than your average new car.
Valve’s New Content Policy for Steam Is a Triumph of Cowardice Over Curation
Valve's complete abdication of the very concept of content curation at any meaningful level flings open the gates to fraud and abuse. But hey. It'll help the company make fewer tough decisions.