AdoredTV has revealed an essential roadmap of Intel’s next-gen mobility platform, including the 15th Gen Arrow Lake CPU family. According to the roadmap, Intel is going all out with its next-gen mobility range, including a powerful GPU and CPU architecture improvement.
Intel appears to be favouring mobile over the desktop with Arrow Lake. While both Arrow Lake-S and Arrow Lake-P CPUs will be available, the company plans to deliver its 15th Gen mobility CPUs first to compete with Apple’s next-generation MacBook 14″ laptops. According to the leaked plan, the first engineering samples will be ready by late 2022 or early 2023, with QS chips shipping out in Q3 2023 and final production starting the next quarter. Finally, in Q4 2023, the CPUs will be ready for RTS (Ready To Ship). As a result, the next-generation Arrow Lake mobility CPU family will debut in late 2023 or early 2024.
The Intel Arrow Lake-P will be a Halo product based on the brand new Lion Cove and Skymont processors, which appear to use TSMC’s N3 manufacturing node. According to Intel’s latest investor day roadmap, the Arrow Lake CPUs will employ ‘Intel 20A’ and ‘TSMC N3’ foundry nodes; therefore, it appears that the Compute tiles, which contain the CPU and GPU, would use an external foundry node from TSMC, while specific SOC/IO IPs will use Intel’s 20A node. Despite rumours that TSMC’s 3nm may miss its launch schedule, and given that the PowerPoint is a few months old, it’s still unclear what the N3 node would be used for in Arrow Lake-P CPUs.
According to other details, since Arrow Lake is the follow-up to 14th Gen Meteor Lake chips, the SOC/IOE-P B step will be 100% reused from the Meteor Lake L/P SKUs. However, the most crucial point is that the Arrow Lake-P CPUs will be configured in 6+8+3 combinations. That’s a 6+8 (P-Core / E-Core) design, bringing the core count to the same 14 cores as Alder Lake-P CPUs, but the iGPU will be based on the Battlemage graphics architecture, with up to 320 Execution Units or 20 Xe-Cores.
Given how clock rates scale on TSMC’s advanced process nodes, that’s a total of 2560 ALUs working at over 2 GHz clock speeds, so we’re in for some remarkable performance from an integrated tile-GPU design. I should also emphasise that we’ve seen GT3 tGPUs with 384 EUs referenced in leaked Intel drivers, so there’s a good chance we’ll see even more high-end Arrow Lake-P designs than what’s included in this (still outdated) roadmap.
It’s also worth mentioning that Arrow Lake-S Desktop CPUs will include up to 40 cores (8 P-Core & 32 E-Cores). So it appears like Intel will go the more efficient way for mobility, at least, as they will only use a portion of the complete core arrangement that desktop chips would get. In addition, Intel has teased a four-chipset architecture for Arrow Lake. The Intel 20A process node will provide a 15% increase in performance per watt while also introducing RibbonFET and PowerVia technology.
Also Read:
Tencent patents its new car applet which can run through browsers
