Iris Xe Max: 5 Things You Need to Know About Intel’s First Discrete GPU
Intel’s Iris Xe Max, debuting on laptops first, is the first discrete graphics solution from the chip giant, but its purpose and feature focus may surprise you.
Rumors have been flying and trickles of official info seeping out about Intel’s first discrete graphics processor. Would it be for desktops? Laptops? Gamers or business go-getters? The chip giant recently provided PCMag with the details in a private briefing.
The solution, named Iris Xe Max, will make its way into a few 11th Generation “Tiger Lake”-based laptops first. (Expect them as 2020 draws to a close.) We have a general overview of Iris Xe Max with more details, but here, I’ve gathered some of the most important points—and what I see as the implications of Xe Max.
We saw plenty of information about Iris Xe Max, but to set the stage, know that it may not end up coming to loads of laptops in practice, at least at the beginning. It aims for the same general discrete-GPU target as Nvidia’s low-end GeForce MX graphics solution, and those chips don’t show up in all that many laptops, relative to those with either integrated (on-CPU) graphics or higher-end discrete graphics.
Specialized content-creation/media-processing laptops, as well as gaming machines, need beefy GPUs—beefier than what the GeForce MX line traditionally serves up. (GeForce GTX, or less often, AMD’s Radeon RX, is the typical minimum class of GPU for gamers.) In contrast, many cheap or portability-first laptops will focus on lower prices and processing power for everyday tasks, meaning GeForce MX or the Iris Xe Max naturally makes the most sense for this niche of in-between laptops. And often, if a laptop OEM is trying to hit a price point, and graphics muscle is not the first order of business, the first thing to get jettisoned is any idea of discrete graphics. You get lower production costs, and it’s less complicated thermally, to use the graphics acceleration found on all modern laptop CPUs—especially as you’re getting them on the chip whether you want ’em or not.
Iris Xe Max will launch in just three laptops to start: the Acer Swift 3X, the Asus VivoBook Flip 14 TP470, and the Dell Inspiron 15 7000 2-in-1. These are all slim, portable laptops, and that’s exactly the type of system in which Iris Xe Max will be utilized. Larger, thicker laptops will, as stated, likely be for a specialized purpose that requires the power of a more potent GPU.
This also makes sense because thin-and-light laptops are also the same types that the first Tiger Lake chips (part of the lower-power U-Series) are present in. Since these two will work best when joining forces (you’ll only find Iris Xe Max in Tiger Lake laptops), expect to predominantly find Iris Xe Max in these kinds of laptops.
2. Surprise! It’s Not Really for Gamers
Intel did share some early gaming graphics test numbers with us. We see some respectable frame rates out of the Iris Xe Max versus integrated graphics, roughly in line with Nvidia’s low-end GeForce MX350 offering. Still, this isn’t being pitched primarily as a gaming boon. That may have been your assumption when you first heard about Intel getting into discrete GPUs, but it makes sense when you consider the rest of the chip giant’s approach to processors.
In our meeting, Intel focused more on how Xe Max would help accelerate media-creation and -editing for pro and prosumer content whizzes. We saw examples like video encoding, boosted by AI, shown as a main benefit of its discrete GPU solution. Intel showed a demo that displayed faster encoding times with Iris Xe Max, and even claimed superiority over high-end Nvidia GeForce cards for some GPU-accelerated tasks, thanks to the new Deep Link AI media encoders baked into Iris Xe Max.
A number of applications will function optimized for Iris Xe Max on day one, such as Handbrake, OBS, XSplit, and more. Popular applications such as Blender, Magix, and more will add support “soon.” In contrast, we saw a couple of gaming frame-rate charts (one is replicated above). In it, Xe Max trades brisk blows with the GeForce MX350, but it doesn’t look like a sea change compared to the integrated Iris Xe on Tiger Lake CPUs already.
The thing is, as I discussed in some of our early testing of Intel’s Iris Xe integrated graphics, the graphics performance of that solution built into the upper-end Tiger Lake chips is already quite decent. ( See my analysis of Intel Iris Xe here.) We haven’t tested Xe Max yet, but layering Xe Max on top of a CPU that already has Iris Xe integrated silicon, from a raw graphics perspective, is likely to amount to replacing “good” with “pretty good.” Not a bad thing, mind you, but to some extent duplicative.
As for gamers, fear not. While this “DG1” (Intel’s codename for “Discrete Graphics 1”) product isn’t aimed at you, Intel is preparing its DG2 discrete graphics solution that’s made directly for gaming. We expect it to release next year for desktops, which leads me to the following point.
3. You’ll See Xe Max in Desktops, Too, But Not as an Upgrade
While it’s certainly a laptop-first technology, expect to see Iris Xe Max make its way to desktops. Intel told us that it is targeting early 2021 for a desktop version. It won’t be a standalone product you can buy to install, build a PC around, or upgrade your existing PC with. It will be sold directly to OEMs and packed into lower-cost consumer PCs.
Affordable home desktops often lack graphics cards, after all, but that doesn’t mean they couldn’t benefit from the muscle of Iris Xe Max for quick photo edits or other media tasks. In a sense, it fills a gap, or adds competition where there is little. Nvidia has traditionally offered low-end GeForce MX desktop cards for designs like value PCs and low-profile systems, where dedicated graphics is desired but thermally tricky to implement. The Iris Xe Max could also be paired with CPUs, like Intel’s special F-series Core CPUs, that lack integrated graphics, or some of AMD’s Ryzen desktop chips, many of which are CPU only, with no integrated GPU. That could make an Xe Max card a cost-saving measure while adding a mild graphics boost over an integrated solution, and gaining the PC OEM the marketing cachet of advertising a “dedicated” card.
Sure, most mainstream and serious gamers will have a more powerful rig with a “proper” GeForce GTX/RTX or AMD Radeon RX GPU, and creative professionals may own an Apple iMac. But many family PCs could stand to gain from what look to be Iris Xe Max’s nontrivial improvements over the integrated graphics common on current desktop chips.
4. It’s All About the ‘Deep Link’ AI, and Intel’s Special Sauce
As with Tiger Lake as a whole, the gains from Iris Xe Max are not entirely due to stuffing in newer, better silicon to improve performance. Intel is leaning on optimizations in background software, under a larger initiative it is calling Deep Link, for these passive gains that act as a special sauce to tie all the advantages together.
The software framework will turn into real-world improvement on the user end, theoretically seamlessly. Deep Link can utilize the power of both the Iris Xe Max GPU and the on-CPU Tiger Lake integrated graphics to boost workloads like video encoding. Interestingly, Intel’s early lab tests have shown that integrated Iris Xe silicon can perform more efficiently on some tasks than Iris Xe Max. So part of Deep Link’s remit will also be to intelligently choose which GPU to harness—the discrete or the integrated—for the appropriate job.
Another feature, Dynamic Power Share, also balances the system’s power delivery toward graphics workloads when needed, pushing as much juice as possible in the direction of a strenuous photo-edit or video-encoding job automatically. This function exists in what Intel calls a more rudimentary form currently, but applying Deep Link learnings should do it even more efficiently. Features like this are baked in across the board to enhance the hardware improvements of Iris Xe, and add something proprietary other than incremental improvements to architecture.
5. For the First Time in a While, We Have Competition in Low-End Graphics
Remember how in 2017 AMD suddenly made the desktop-CPU space wildly competitive with the introduction of its first-generation, many-core Ryzen CPUs? Iris Xe Max could be a parallel, if more minor, tremor in the less-trafficked world of low-end mobile graphics for laptops.
Merely having real graphics competition to speak about at this power tier is essentially a new phenomenon in itself. Previously, Intel’s HD Graphics and UHD Graphics integrated solutions were barely useful for gaming, AMD’s APU-based Radeon ones were reliably a tad better, and Nvidia’s GeForce MX line was the closest step to real gaming power short of a dedicated GTX chip.
Now, we’re debating the merits of multiple options. Iris Xe itself is moving the baseline way up for integrated, and Iris Xe Max is both working in conjunction with it and offering even better performance. Nvidia’s latest GeForce MX GPU is still in the mix, and so we have an actual debate about which is best, rather than a default choice for “tweener” graphics.
Each has its own advantages (Iris Xe Max seems to bring the most holistic boosts), and their presence should push the other along to dominate this low-end space. When was the last time Nvidia had to fight for this ground, after all? Now GeForce MX is getting pressed on one side from better-than-ever integrated solutions in the Iris Xe, and a new challenger at its level on discrete in Iris Xe Max.
The possible outcome: Real-world applications will see meaningful improvements, which is far more than we could say before, and basic gaming will be opened up to new users thanks to lower price points. (Many more of the implications of integrated Iris Xe itself were covered in a different piece I referenced earlier. Check that out for more perspective on the “junior” Xe.) This should benefit consumers, as these options battle to make a case as the best value for affordable, portable systems. Bring on the Xe, I say.
Originally published at https://www.pcmag.com.
