Intel has taken the best from AMD and the best from Nvidia is gearing up to make a splash in the discrete graphics card market. And I’m not just talking about the number of engineers, designers, and marketers Intel has poached from the big two in GPU gaming. From the design of new Alchemist architecture itself, to the software, to the scaling solution, it’s clear that Intel is learning from the best.
That’s not to say Intel isn’t bringing its own style to the party because it launches its first mobile Arc GPUsbut when trying to navigate a market dominated by just two players for decades, you have to be careful about what works and not reinvent the wheel.
Before you get too excited about a third player in the GPU market, we’re still not talking about Intel’s discrete desktop graphics cards yet. Intel just announced its Arc 3 GPUs built into thin and light laptops starting in early April. Then, at the beginning of the summer, it will be the Arc 5 and Arc 7 GPUs; the discrete silicon that powers all-Intel gaming laptops.
However, there is nothing but a tease on the desktop versions, and so the wait continues… But it should be clarified that the basic design will be identical from the lowest laptop GPU to the most powerful desktop chip.
So up first we have Arc 3 in A350M and A370M forms, with six and eight Xe cores respectively (think SM in Nvidia parlance). Next we’ll have the biggest wood, the Arc 5 A550M with 16 Xe cores, and then a pair of high-end Arc 7 GPUs – A730M and A770M – with 24 and 32 Xe cores.
Intel builds these graphics cards from a pair of mobile A-series SoCs, the ACM-G11 with up to 8 Xe cores for Arc 3 and the ACM-G10 with up to 32 Xe cores for the Arc 5 and 7 cards.
We’ve talked before about these designs, from the time they were first leaked many moons ago, in terms of number of execution units (EUs) of 96, 128, 256, 384 and 512 EUs. The older UE design worked with Intel’s older graphics chips, but they’re now known as Xe Vector Engines (XVE) and look a lot like AMD’s RDNA-based dual-workgroup design. .
But attached to each of the XVEs is an XMX Matrix Engine that looks a lot like a mini Nvidia Tensor Core, and each Xe-core comes with a dedicated ray tracing unit. Interestingly, even the lowest GPU will come with ray tracing hardware, and it will be interesting, given the traditional requirements of RT features, how these are used.
The fact that these two best mobile graphics cards come with 12GB and 16GB of GDDR6 memory, attached to 192-bit and 256-bit memory buses, also shows just how serious Intel is about making high-end GPUs for gaming. mobile devices.
While waiting for desktop releases, you can expect some parity between the designs of these mobile chips and their non-mobile siblings. I would expect the core count to line up (although we probably won’t see a six-core Xe desktop chip), but they will come with higher clock speeds and more extreme power demands Consequently.
Again, showing how Intel learned from the other two, the Xe Super Sampling (XeSS) uses Arc’s XMX AI engine to power the temporal upscaler, much like Nvidia’s DLSS uses Tensor Cores of Ampere.
In terms of real-world performance, we’re still a bit in the dark on high-end GPUs, but Intel has at least shared the level of mainstream gaming framerates you can expect from the Arc A370M GPU. Basically, you can expect 1080p gaming from your new ultra-mobile laptop at a consistent 60fps. How this compares to AMD’s latest integrated graphics, we’ll have to wait and see until we get some comparison laptops for testing. But at first glance, the 8 Xe-core chip can outperform the 680M.
Which should also put it above the Nvidia MX 450 at the lower end of Nvidia’s GPU stack.
It’s an exciting time, to finally be on the verge of having a third entrant in the two-horse GPU race. And, while low-end mobile graphics chips aren’t necessarily the ones we really want to put our tech gloves on, they can be a solid foundation on which Intel can build. And I can’t wait to test them out when I finally get my hands on them next month.