Intel has reportedly deferred its orders for 3 nm wafers with TSMC, sources in PC makers tell Taiwan-based industry observer DigiTimes. Built on the TSMC N3 node, the wafers were supposed to power the Graphics tiles (containing the iGPU), of the upcoming "Arrow Lake" processors, which were originally on course for a 2024 release. The DigiTimes report detailing this development says that Intel's 3 nm wafer orders have been deferred to Q4-2024, which would realistically mean a 2025 launch for whatever product was designed to use 3 nm tiles. Advance orders for next-gen wafers by high-volume clients such as Intel, are usually placed several quarters in advance, so the foundry could suitably scale up its capacity.

Intel's 2023 roadmap for the desktop processor segment sees the company flesh out its 13th Gen Core "Raptor Lake" desktop family with 65 W (locked) SKUs, and the new i9-13900KS flagship; followed by a new lineup of processors under the "Raptor Lake Refresh" family, due for Q3-2023, with no mentions of a desktop "Meteor Lake" processor in the year. It turns out that "Raptor Lake Refresh" is being designed to fill in for these (i.e. there won't be any "Meteor Lake" desktop chips). This, according to OneRaichu, a reliable source with Intel leaks.

"Meteor Lake" is Intel's first client processor to fully incorporate the company's IDM 2.0 product development strategy of disintegrating the processor into multiple chiplets built on various foundry nodes based on design needs; and combining them onto a single package with a high-performance interconnect. "Meteor Lake" has just one problem and that is CPU core-counts, with rumors pointing to 6P+16E (6 performance cores + 16 efficiency cores) being the maximum core-count possible, something Intel probably feels won't be competitive in the desktop segment against AMD, which will probably have a lineup of "Zen 4" X3D processors out by Q3-2023, with up to 16 P-cores. The company will, however, give "Meteor Lake" a sizable launch in the various mobile segments.

Intel as part of its development process with industry partners and OEMs, allegedly released technical documents in a bare URL that's worded to confirm that its next desktop processor socket will in fact be the LGA1851. We've had some idea since June 2021 that LGA1851 will succeed LGA1700, but this can be taken as a confirmation. Although with a higher pin-count, the LGA1851 package will be physically of an identical size to LGA1700, with mostly identical socket mechanism, so the new socket could maintain cooler compatibility with its predecessor. The additional 151 pins come from shrinking the "courtyard" (the region of the land grid in the center that lacks pins and instead has some electrical ancillaries).

The new Socket LGA1851 platform is expected to power Intel's "Meteor Lake-S" and "Arrow Lake-S" microarchitectures. Whether "Meteor Lake-S" gets the 14th Gen Core branding is a whole different question. Leaked benchmarks suggest that 2023 will be a rather slow year from Intel in the area of desktop processors, and that toward Q3-2023, the company will release the so-called "Raptor Lake Refresh" processors. These chips are likely built on the same LGA1700 package, and as we've seen from "Coffee Lake Refresh," could warrant a new generational branding to 14th Gen Core (as CFL Refresh formed the 9th Gen Core). Intel could increase clock-speeds, E-core counts, and other process/packaging-level innovations to segment these chips apart from existing 13th Gen Core "Raptor Lake." LGA1851 processors like "Meteor Lake" could debut chiplets for Intel, as these have their CPU cores, iGPU, memory-controllers, and uncore components, spread apart on chiplets built on various foundry nodes.

Keeping up with the cadence of two generations of desktop processors per socket, Intel will turn the page of the current LGA1700, with the introduction of the new Socket LGA1851. The processor package will likely have the same dimensions as LGA1700, and the two sockets may share cooler compatibility. The first processor microarchitecture to debut on LGA1851 will be the 14th Gen Core "Meteor Lake-S." These chips will feature a generationally lower CPU core-count compared to "Raptor Lake," but significantly bump the IPC on both the P-cores and E-cores.

"Raptor Lake" is Intel's final monolithic silicon client processor before the company pivots to chiplets built on various foundry nodes, as part of its IDM 2.0 strategy. The client-desktop version of "Meteor Lake," dubbed "Meteor Lake-S," will have a maximum CPU core configuration of 6P+16E (that's 6 performance cores with 16 efficiency cores). The chip has 6 "Redwood Cove" P-cores, and 16 "Crestmont" E-cores. Both of these are expected to receive IPC uplifts, such that the processor will end up faster (and hopefully more efficient) than the top "Raptor Lake-S" part. Particularly, it should be able to overcome the deficit of 2 P-cores.

Intel's next-generation desktop processor socket will be the LGA1851. Leaked documents point to the next-generation socket being of identical dimensions to the current LGA1700, despite the higher pin-count, which could indicate cooler compatibility between the two sockets, much in the same way as the LGA1200 retained cooler-compatibility with prior Intel sockets tracing all the way back to the LGA1156. The current LGA1700 will service only two generations of Intel Core, the 12th Generation "Alder Lake," and the next-gen "Raptor Lake" due for later this year. "Raptor Lake" will be Intel's last desktop processor built on a monolithic silicon, as the company transitions to multi-chip modules.

Intel Socket LGA1851 will debut with the 14th Gen Core "Meteor Lake" processors due for late-2023 or 2024; and will hold out until the 15th Gen "Arrow Lake." Since "Meteor Lake" is a 3D-stacked MCM with a base tile stacked below logic tiles; the company is making adjustments to the IHS thickness to end up with an identical package thickness to the LGA1700, which would be key to cooler-compatibility, besides the socket's physical dimensions. Intel probably added pin-count to the LGA1851 by eating into the "courtyard" (the central gap in the land-grid), because the company states that the pin-pitch hasn't changed from LGA1700.

Hot Chips 34, the upcoming semiconductor conference from Sunday, August 21 to Tuesday, August 23, 2022, will feature many significant contributions from folks like Intel, AMD, Tesla, and NVIDIA. Today, thanks to Intel's registration at the event, we discovered that the company would present its work on Meteor Lake and Arrow Lake processors with the novel Foveros 3D packaging. The all-virtual presentation from Intel will include talks about Ponte Vecchio GPU and its architecture, system, and software; Meteorlake and Arrowlake 3D Client Architecture Platform with Foveros; and some Xeon D and FPGA presentations. You can see the official website here for a complete list of upcoming talks.

As a little reminder, Meteor Lake is supposed to arrive next year, replacing the upcoming Raptor Lake design, and it has already ahs been pictured, which you can see below. The presentation will be recorded and all content posted on Hot Chips's website for non-attendees to catch up on.

Le Comptoir du Hardware scored a die-shot of a 2P+8E core variant of the "Meteor Lake" compute tile, and Locuza annotated it. "Meteor Lake" will be Intel's first processor to implement the company's IDM 2.0 strategy to the fullest. The processor is a multi-chip module of various tiles (chiplets), each with a certain function, sitting on die made on a silicon fabrication node most suitable to that function. Under this strategy, for example, if Intel's chip-designers calculate that the iGPU will be the most power-hungry component on the processor, followed by the CPU cores, the graphics tile will be built on a more advanced process than the compute tile. Intel's "Meteor Lake" and "Arrow Lake" processors will implement chiplets built on the Intel 4, TSMC N3, and Intel 20A fabrication nodes, each with unique power and transistor-density characteristics. Learn more about the "Meteor Lake" MCM in our older article.

The 2P+8E (2 performance cores + 8 efficiency cores) compute tile is one among many variants of compute tiles Intel will develop for the various SKUs making up the next-generation Core mobile processor series. The die is annotated with the two large "Redwood Cove" P-cores and their cache slices taking up about 35% of the die area; and the two "Crestmount" E-core clusters (each with 4 E-cores), and their cache slices, taking up the rest. The two P-cores and two E-core clusters are interconnected by a Ring Bus, and share an L3 cache. The size of each L3 cache slice is either 2.5 MB or 3 MB. At 2.5 MB, the total L3 cache will be 10 MB, and at 3 MB, it will be 12 MB. As with all past generations, the L3 cache is fully accessible by all CPU cores in the compute tile.

Intel is designing a "Halo" SKU of a future generation of mobile processors with a goal to match Apple's in-house silicon of the time. Slated for tape-out some time in 2023, with mass-production expected in 2024, the 15th Generation Core "Arrow Lake-P Halo" processor is being designed specifically to compete with Apple's "premium 14-inch laptop" (presumably the MacBook Pro) that the company could have around 2024, based on an in-house Apple silicon. This is to essentially tell its notebook partners that they will have an SoC capable of making their devices in the class truly competitive. Apple relies on a highly scaled out Arm-based SoC based on in-house IP blocks, with a software that's closely optimized for it. Intel's effort appears to chase down its performance and efficiency.

The Core "Arrow Lake" microarchitecture succeeds the 14th Gen "Meteor Lake." It is a multi-chip module (MCM) of three distinct dies built on different fabrication nodes, in line with the company's IDM 2.0 strategy. These nodes are Intel 4 (comparable to TSMC N7 or N6), Intel 20A (comparable to TSMC N5), and an "external" 3 nm-class node that's just the TSMC N3. The compute tile, or the die which houses the CPU cores, combines a hybrid CPU setup of 6 P-cores, and 8 E-cores. The performance cores are likely successors of the "Redwood Cove" P-cores powering the "Meteor Lake" compute tiles. Intel appears to be using one kind of E-cores across two generations (eg: Gracemont across Alder Lake and Raptor Lake). If this is any indication, Arrow Lake could continue to use "Crestmont" E-cores. Things get interesting with the Graphics tile.

Intel's upcoming "Meteor Lake" and "Arrow Lake" client mobile processors introduce an interesting twist to the chiplet concept. Earlier represented in vague-looking IP blocks, new artistic impressions of the chip put out by Intel shed light on a 3-die approach not unlike the Ryzen "Vermeer" MCM that has up to two CPU core dies (CCDs) talking to a cIOD (client IO die), which handles all the SoC connectivity; Intel's design has one major difference, and that's integrated graphics. Apparently, Intel's MCM uses a GPU die sitting next to the CPU core die, and the I/O (SoC) die. Intel likes to call its chiplets "tiles," and so we'll go with that.

The Graphics tile, CPU tile, and the SoC or I/O tile, are built on three different silicon fabrication process nodes based on the degree of need for the newer process node. The nodes used are Intel 4 (optically 7 nm EUV, but with characteristics of a 5 nm-class node); Intel 20A (characteristics of 2 nm), and external TSMC N3 (3 nm) node. At this point we don't know which tile gets what. From the looks of it, the CPU tile has a hybrid CPU core architecture made up of "Redwood Cove" P-cores, and "Crestmont" E-core clusters.