Latest Intel Arrow Lake Leaks Point To Interesting Design Choices
Latest Intel Arrow Lake Leaks Point To Interesting Design Choices
Intel’s Arrow Lake architecture is set to succeed the 14th Gen “Raptor Lake Refresh” processors later this year, and more information has emerged from leaker Golden Pig Upgrade (金猪升级包) (via Videocardz). As far as leaks go, this is a particularly interesting one, as there are many changes made to fundamental design choices.
The post on Bilibili started with the leaker stating, “(I have) no idea why a processor set to launch in slightly more than half a year is still getting many wildly inaccurate leaks…” The leaker then clears things up with what’s known. Starting with the naming – the leaker says it’s “impossible” for Arrow Lake to be referred to as 15th Gen Core, and instead predicts that “Core Ultra (Series 2)” will be the name when it launches.
A second piece of information indicates that Arrow Lake will not feature hyper-threading and LPE-core (battery runtime isn’t a concern on desktops), but it retains NPU from the Meteor Lake architecture. As a result, the AI performance will mirror the laptop architecture.
This information matches the previous leak that indicates Arrow Lake will ditch the hyper-threading feature. What happens to multi-core performance, then? In a reply, the leaker alleges that there are “no concerns for now” on the possibility of performance regression.
For onboard graphics, all chips will be upgraded to 4 Xe cores, though it won’t be referred to as “Intel Arc Graphics”. Since the name is only given to onboard graphics with 7 or more Xe cores, “Intel Graphics” will be what you’ll be seeing in Task Manager. Also, DDR4 is no longer supported.
Golden Pig Upgrade also touches on the mobile segment expected to spin off from the desktop lineup. The leaker notes that laptop chips will feature multiple ‘Tiles’, of which the “leading edge” variants are set to be manufactured by TSMC, whereas Intel’s own 20A process will only be responsible for 6+8 core configuration (or its binned variants) under the Arrow Lake-S lineup, which likely includes the Core Ultra 5 non-K and lower.
It’s worth noting here that Intel’s 20A process is several nodes more advanced than the current node used by Intel’s current desktop chips – which is Intel 7 (and was originally called 10nm Enhanced SuperFin before the renaming). The chipmaker will be skipping Intel 4 (used for Meteor Lake) and Intel 3 altogether, although this big of a jump likely leaves the question of the node’s yield status.
Additionally, the leaker also noted that Intel is still unable to design a PCH-less solution for its ARL-S and HX-series chips, unlike AMD’s processors. Meaning, that gaming laptops sporting Intel’s upcoming HX-series still require a discrete chipset to operate various downstream connections, which AMD’s HX-series equivalents no longer require.
Pokdepinion: That’s a huge amount of information to take in. I wonder how Intel deals with the potential loss of multicore performance now that HT is gone?
