Intel Unveils 14th Gen Meteor Lake: Next Gen Core Architectures on Intel 4 Process Node

On the first day of its Innovation Event, Intel unveiled its 14th Gen Meteor Lake mobility processors featuring brand new core architectures and its 4nm (Intel 4) process node. The P-core has been upgraded to Redwood Cove, promising “improved performance efficiency”, increased inter-core bandwidth (new interconnect), and a well-trained Thread Director to properly assign workloads between the two core clusters.

A high-level look at the block diagrams shows little to no difference between Redwood and Golden Cove. That said, a lot of the important details, such as ROB size, register width, cache, and scheduler bandwidth, are still unknown, so we’ll have to wait a little more before diving into the new microarchitecture. What Intel did say about these chips is that they’ll leverage advanced 3D packaging (Foveros 2nd Gen) with a separate iGPU and SoC tile.

The same can be said about Crestmont, which gets a wider instruction allocation buffer (5->6 wide) and support for VNNI (Vector Neural Network Instructions), complemented by twice as many AVX2 VNNI ports.

Thread Director was introduced with Alder Lake to address the shortcomings of Windows’ Scheduler (or perhaps complement it). Intel is promising substantial improvements to Thread Director with Meteor Lake. With Alder/Raptor Lake, the TR prioritized the P-cores by assigning them the higher QoS work, leaving the remaining lower QoS work for the E-core. The latter was continually shifted to the former as the P-cores returned to idle.

With Meteor Lake, the workload is first contained on the E-cores on the SoC, failing which, they are shifted to the E-cores on the compute die. If even more performance is required, the threads are moved to the P-cores. This strategy prioritizes power efficiency, something that was sorely lacking on Alder and Raptor Lake.

And now, for the other major upgrade, the Intel 4 process node. This 4nm-class node doubles the area scaling for high-performance logic libraries versus the existing Intel 7 node. It also offers 20% more power efficiency compared to its predecessor with EUV lithography to help raise the production yields. Of course, there’s also a MIM (Metal Insulator Metal) Capacitor, which improves power delivery on high-end chips.


Computer hardware enthusiast, PC gamer, and almost an engineer. Former co-founder of Techquila (2017-2019), a fairly successful tech outlet. Been working on Hardware Times since 2019, an outlet dedicated to computer hardware and its applications.
Back to top button