Right now is when Intel does its third-quarter 2021 monetary disclosures, and there’s one little tidbit within the earnings presentation about its upcoming new discrete GPU choices. The earnings are often an opportunity to wave the flag of innovation about what’s to return, and this time round Intel is confirming that its first-generation discrete graphics with the Xe-HPG structure will probably be on cabinets in Q1 2022.
Intel has slowly been disclosing the options for its discrete gaming graphics choices. Earlier this 12 months, the corporate introduced the branding for its next-gen graphics, referred to as Arc, and with that the primary 4 generations of merchandise: Alchemist, Battlemage, Celestial, and Druid. It’s simple to see that we’re going ABCD right here. Technically at that disclosure, in August 2021, Intel did state that Alchemist will probably be coming in Q1, the reaffirmation of the date as we speak within the monetary disclosures signifies that they’re staying as near this date as potential.
Intel has beforehand confirmed that Alchemist will probably be totally DirectX 12 Final compliant – which means that alongside RT, it is going to supply variable-rate shading, mesh shaders, and sampler suggestions. It will make it comparable in core graphics options to current-generation AMD and NVIDIA {hardware}. Though it has taken a number of years now to return to fruition, Intel has made it clear for some time now that the corporate has supposed to develop into a viable third participant within the discrete graphics house. Intel’s odyssey, as earlier advertising efforts have dubbed it, has been pushed primarily by growing the Xe household of GPU microarchitectures, in addition to the GPUs primarily based on these architectures. Xe-LP was the primary out the door final 12 months, as a part of the Tiger Lake household of CPUs and the DG1 discrete GPU. Different Xe household architectures embody Xe-HP for servers and Xe-HPC for supercomputers and different high-performance compute environments.
The basic constructing block of Alchemist is the Xe Core. For manufacturing, Intel is popping to TSMC’s N6 course of to do it. Given Intel’s Q1’22 launch timeframe, Intel’s Alchemist GPUs will virtually definitely be essentially the most superior shopper GPUs in the marketplace with respect to manufacturing expertise. Alchemist will probably be going up in opposition to AMD’s Navi 2x chips constructed on N7, and NVIDIA’s Ampere GA10x chips constructed on Samsung 8LPP. That stated, as AMD can attest to, there’s extra to being aggressive within the shopper GPU market than simply having a greater course of node. At the side of using TSMC’s N6 course of, Intel is reporting that they’ve improved each their energy effectivity (performance-per-watt) and their clockspeeds at a given voltage by 50% in comparison with Xe-LP. Word that that is the sum whole of all of their enhancements – course of, logic, circuit, and structure – so it’s not clear how a lot of this comes from the bounce to TSMC N6 from Intel 10SF, and the way a lot comes from different optimizations.
Precisely what efficiency stage and pricing Intel will probably be pitching its discrete graphics to is at the moment unknown. The Q1 launch window places CES (held the primary week of January) as a great spot to say one thing extra.