Introduction

Intel today announced their 12th Generation Core "Alder Lake" desktop processors, debuting with six "unlocked" models aimed at gamers and PC enthusiasts. Intel claims that these will be "the world's best gaming processors," marking a giant leap in performance for content-creators while being the most rewarding for enthusiasts and overclockers.

Today, we are allowed to share first details with you about the fascinating new hybrid architecture that's making its debut to the desktop segment with these processors—the biggest change to the x86 ecosystem since the introduction of x64—and many other innovations they bring. The processors are open for pre-orders starting today and will be generally available from November 4 onward. For performance numbers from our reviews, you'll have to check back with us on that same date as that's when the Intel review embargo lifts.



In a recent interview, Intel CEO Pat Gelsinger claimed that Intel is back with performance leadership, and that its competitor AMD, which currently has the lead, is "over." This underscores the importance Intel attaches to Alder Lake. AMD's latest Ryzen 5000 "Zen 3" processors comprehensively beat Intel both in single and multi-threaded performance. The red team has been tormenting Intel ever since Ryzen's debut in 2017, which is partially Intel's own fault. Intel was comfortable selling 4-core/8-thread processors to the mainstream segment, taking their time with innovation. With the company's transition from 14 nm to 10 nm slowing down, the company found itself stuck with a flat-core IPC growth rate since 2016, and hence had to dial up core/thread counts with each generation since 8th Gen Core.

Intel did achieve 10 nm mass-production as far back as 2019, but market demand surged, causing chip shortages, and Intel probably decided to allocate all its foundry capacity to making mobile chips that could benefit from the 10 nm node, leaving desktop with 14 nm until now. The new 12th Gen Core Alder Lake processors are built on the new "Intel 7" process, which is technically 10 nm Enhanced SuperFin rebadged. Intel isn't totally off as that process offers comparable transistor-density and power characteristics to 7 nm-class DUV nodes from TSMC or Samsung.

The switch to Intel 7 enabled the company to dial up transistor counts, and deploy its latest "Golden Cove" high-performance cores, but it faced a problem. Intel does not [yet] believe in the CPU chiplet approach to building processors with large CPU core-counts that AMD uses. Given the target die-area of Alder Lake, it could just about deploy 10 performance cores, which come in short compared to AMD's offerings, not only psychologically. Intel's engineers came up with an innovative solution: adding two quad-core "Gracemont" low-power core clusters in place of what would've been the 9th and 10th CPU core. Alder Lake is now a 16 core processor, with 8 "performance cores", and 8 "efficiency cores." Seems like a lazy way to catch up to AMD in the core-count game? This is where Intel's smarts really come in.


Intel figured out that its Golden Cove P-cores already give it a 28% IPC gain over the "Skylake" core (about 15–20% over AMD's Zen 3). If it could just design its Gracemont E cores to be reasonably fast (think within 10% of Skylake), the net performance of the processor would be similar to a 16-core Ryzen. Games don't need 16 cores and would benefit from the sheer weight of the IPC gain of the P-cores, while multi-threaded productivity tasks would benefit from the cumulative performance of both core types. When neither gaming nor dealing with heavy productivity workloads, the processor could send the P-cores to sleep, operating them at very low power, and let the E cores deal with the workloads, thereby giving consumers a leap in efficiency. This is all easier said than done from the perspective of software.

In this article, we'll dive a little deeper into the inner workings of Alder Lake. Our performance reviews of these processors will go live on November 4.

The 12th Generation Core Lineup

Intel is launching its 12th Gen Core desktop processors in two stages. Today's launch only includes the Unlocked "K" and "KF" SKUs along with the top Z690 motherboard chipset, so gamers and PC enthusiasts have something to pick up for this holiday. The company will augment the larger half of this lineup, the numerous non-K variants, along with more affordable motherboards based on mid-tier chipsets some time in Q1-2022.


Leading the pack are the Core i9-12900K and i9-12900KF, followed by the Core i7-12700K and i7-12700KF, and the Core i5-12600K and i5-12600KF in the middle tier. The introduction of the E-core clusters gives Intel a new dimension for segmenting its lineup, particularly between the Core i9 and Core i7 SKUs. The core-counts are represented as P+E (performance + efficiency cores). The performance cores support HyperThreading, while the efficiency cores lack it.

The Core i9-12900K and i9-12900KF are 8+8 core SKUs, fully unlocking all CPU cores physically present on the silicon. The Core i7-12700K and i7-12700KF are 8+4 core SKUs, in which one of the two E-core clusters is disabled. The Core i5-12600K and i5-12600KF are 6+4 core processors, in which two P-cores and one E-core cluster is disabled. In addition to cores, Intel uses the shared L3 cache size for segmentation, too. The Core i9 parts feature the full 30 MB of L3 cache physically present. The Core i7 ones feature 25 MB. The Core i5 parts feature 20 MB. The Core i9 and Core i7 series chips support Turbo Boost Max 3.0, while the Core i5 SKUs lack it.

Intel introduced Thermal Velocity Boost a few generations ago. It's not included on any Alder Lake SKU—we confirmed that directly with Intel. What's also been removed is support for AVX512.

Prices of these processors remains roughly as steep as the 11th Gen "Rocket Lake." Intel announces MSRPs per chip in 1,000-unit tray quantities. The i9-12900K is priced at USD $589, while the i9-12900KF goes for $564. The Core i7-12700K is priced at $409, with the i7-12700KF going for $384. The Core i5-12600K goes for $289, and the i5-12600KF for $264.

You'll notice that the K and KF SKUs are consistently priced $25 apart. This is the value Intel attaches to the UHD 770 integrated graphics (iGPU), which the KF SKUs lack. The KF SKUs might appeal to gamers with discrete graphics cards, who may not bother with the iGPU. The money saved could be spent toward getting an aftermarket cooling solution, as none of the six SKUs being launched today feature a stock cooler inside the box. This is particularly disappointing because these processors herald a new CPU socket, the LGA1700, and its cooler mounting design as well as IHS coverage requirements differ from older LGA1200 or LGA115x sockets. Scouting for a compatible cooler or upgrades to your existing ones will be as difficult than it was for AMD users in 2017. Some motherboard vendors have decided to include mounting holes for both the current LGA115x/LGA1200 coolers and new LGA1700, so your old coolers might have a little bit more life left in them.

The Alder Lake-S Silicon

As we mentioned earlier, Intel still does not subscribe to the modular CPU core chiplet approach to build high core-count processors used by AMD. Its hybrid architecture approach, though, could be described as being just as innovative. The "Alder Lake-S" silicon is a monolithic die fabricated on the Intel 7 process, known earlier as 10 nm Enhanced SuperFin. The silicon has essentially the same layout as Intel client processor dies from the past, with a centralized last-level cache (LLC, or L3 cache) and a bi-directional ringbus connecting the various blocks (parts of the processor).

The most important blocks are the CPU cores, and there are 16 of them on Alder Lake-S. Eight of these are the larger Golden Cove P-cores, and eight of these are Gracemont E cores. Each of the eight P-cores has its dedicated ring-stop. Groups of four E cores make up an E-core cluster. There are two such E-core clusters on Alder Lake-S, each with a ring-stop. From what we can tell, while you can disable individual P-cores available on your SKU, you can only disable E cores on a per-cluster basis. You also must have at least one P-core active—so you can't run the CPU with only E cores. You can disable all E cores for a pure P-core processor, though.

The Alder Lake-S silicon features 30 MB of shared L3 cache that's accessible by all components on the silicon, including the P-cores, E-core clusters, iGPU, memory controllers, PCIe root-complex, and other uncore components. Each of the eight P-cores has a dedicated 1.25 MB of L2 cache. Each E-core Cluster has a 2 MB L2 cache that's shared among the four E cores. We'll describe the L1 caches that deals with each of the two CPU core types on the next page.