Intel has announced its new brand, Arc, under which the iconic chipmaker will—for the first time—make discrete high-performance consumer GPUs. Arc graphics cards will focus largely on gamers and content creators which is to say, Intel is going after NVIDIA and even AMD in what may seem like uncharted territory to begin with.
The first laptops and desktops based on first-generation Intel Arc GPUs—codename Alchemist—are slated to release in early 2022. Although Intel has demonstrated few aspects like variable rate shading, video upscaling, AI-accelerated game supersampling and mesh shading piquing lot of interest from the industry—and the gaming community—more granular details including those related to configurations and pricing are yet to be revealed. Those will be shared closer to launch.
In the meantime, Financial Express Online decided to catch up with Intel’s Prakash Mallya to understand some of the nitty-gritties of Alchemist. More importantly, why is Intel making discrete GPUs now. Excerpts.
FE: The world is curious about Alchemist because it’s something completely new, completely different from what you do conventionally. We won’t get into the discussion if NVIDIA (and AMD) should be worried, at this point, but could we just start with Alchemist? Why is Intel making GPUs?
Prakash Mallya: We are operating in times when every usage model and every industry vertical is going to be either AI, network, 5G or Edge-oriented. The cloud workload and AI workload to Edge workload and network workload and all the different variants within them are going to be very different. That will demand you to create architectures which are best suited for those needs.
When you look at the overall Xe architecture—of which Alchemist is a part—it goes from low power to really high-performance computing and AI in Ponte Vecchio, and it also does enthusiasts and gaming. The Xe architecture in itself has got so many variants, because of the variety of workloads.
Specifically, on Xe HPG, which is Alchemist and the future generations that we (have) revealed, it is a specific architecture that combines in client situations and gives very unique solutions, especially for enthusiasts and gamers. It’s software oriented, it’s architected for software first. It supports Direct X12 Ultimate as well as a standard and that makes it very easy for an ecosystem to build on top of it. Plus, it provides a great experience on performance.
So, you would see when we launch this early part of next year with our customers building on top of it and our ecosystem—the work that we do on software, enabling the software before the product launches—will help Alchemist really shine in the usages that it is built for, which is gaming, content creation and others.
FE: A lot of its technical details are still under the wraps. Not a lot of information is out there; will you be willing to share a little bit of insight into what we are looking at?
Prakash Mallya: It is a brand-new product line for us. The target segment that Alchemist is focused on—which is gaming, enthusiasts and content creators—demands performance and capabilities which you don’t see in other consumer client usages. So, I would say that wait for more details. It is an exciting product that we are all really thrilled about. We are committed towards not only Alchemist, but the more interesting piece is the future generation of roadmaps. So, we are not delivering ‘a’ product, we are delivering a set of products over multiple years that shows our commitment to this category. Because that category of customers demands horsepower, demand performance, and we will give it to them.
Two features that you should really look forward to, is our super sampling method—XeSS—which is going to enable high resolution gameplay without costing frame rates and then there’s also our support for ray tracing. So, we are bringing industry-leading features implemented via a combination of hardware and software. We are also promising Day Zero drivers for games. So, essentially our entire approach to Alchemist is putting gamers and their needs first.
FE: Xe is a very popular product and it’s brought in a lot of capabilities especially to thin and light laptops. Where do you see it going next with you getting into the discrete GPU space?
Prakash Mallya: The Xe architecture is a fully scalable graphics architecture that will serve many markets. We’ve built Intel’s Xe architecture to scale the breadth of GPU spectrum from the efficiency and mobility-focused Xe LP architecture, the gaming and enthusiast-focused Xe HPG, right up to the AI and datacentre-focused Xe HPC architecture. The reason why we are taking so many different approaches to the Xe architecture is because of the evolving needs of consumers. For the longest period of time, we’ve had integrated graphics in our portfolio and that caters to majority of usages—in the mainstream—and to a little bit of performance needs also. When people are pushing for more performance, they need discrete GPUs. That’s where the discrete graphics orientation came about.
Now, when you see our roadmap from integrated graphics to discrete graphics as well as the future generation of discrete graphics, we have a very wide portfolio, and it also goes from a wide range of form factors that it can (enable). So, that gives you the flexibility. One size does not fit all holds true even for graphics usages and graphics-oriented users.
FE: What does the future look like for integrated—Xe—graphics?
Prakash Mallya: The integrated graphics is a definite piece of technology that we will continue to invest in because we believe in catering to a variety of usages, and this is a great example of adding more to that portfolio so that we are really sharp on indexing towards different usages with unique products.
Also Read | Rocket Lake-S deep dive: Intel on broken embargoes, benchmark scores, walled gardens and power of PC
