Intel may be on the verge of a comeback, though it will take some time.
Essentials
- Intel’s next-generation Meteor Lake chips, which should arrive in laptops in 2024, are an exciting development that will pave the way for CPUs in various fields.
- Meteor Lake will use Intel’s new tile technology, with four tiles dedicated to specific features: a tile with CPU cores, an SoC tile with an interface and an AI processor, a GPU tile with graphics cores, and a tile with IO with necessary connections.
- Intel’s strategy aims to counter TSMC by releasing Intel 4 (twice the density of Intel 7) and Intel 3, followed by the launch of 20A in 2025, which the company believes will bring success after a difficult decade.
At this year’s Intel Innovation event, Intel released key (although not unexpected) details of the Meteor Lake chips, which are coming to next-generation laptops in 2024. Although Intel is not bringing this to the desktop, this. is one of the most exciting generations of Intel hardware in recent years. In fact, Meteor Lake is the pipe cleaner that will make way for Intel’s CPUs in several sectors, from laptops to desktops to datacenters to AI.
The Meteor Lake specification is Intel’s new system for CPUs
We’ve known about Meteor Lake for a while now: it will be built using Intel technology, which is very similar to AMD chipsets. However, the main difference between Intel and AMD tiles is that Intel provides each tile with a specific function, and Meteor Lake will come with four tiles: Compute tile with CPU cores; an SoC tile with display connectivity, wireless technology, and an AI processor called a Neural Processing Unit (or NPU); GPU tiles with graphics cores; and an IO tile with the necessary PCIe and Thunderbolt connectors.
Although Intel probably has many different chips in the works today, the Meteor Lake chip Intel has revealed so far will have a base configuration of 6-performance cores and 8-efficiency cores, two less efficient cores than the high-end Raptor Lake chip. Core i9-13900K. However, since Meteor Lake will appear only on computers, the lack of two P-cores will not be a big problem, since laptops have very little power. Meteor Lake’s Compute tile is built on the Intel 4 node, formerly known as Intel 7nm.
The P-cores and E-cores are also new, with increased multi-core performance and increased core bandwidth, in the case of the P-cores. For E-cores, Intel is also offering updates to the instructions on the process (IPC), and they also come with the Thread Director control technology. This should allow the processor to make intelligent decisions about which cores to use at certain times, providing better performance and power consumption when needed.
We don’t know much about the Xe-LPG architecture that will make up the Meteor Lake chip for Intel Arc, but Intel says it has two times the current Xe-LP architecture, which would mean twice as much. Despite the clock speed boost, the graphics chip is more efficient than what’s in Intel CPUs, and AI helped get it there. Intel also says it supports features like ray tracing and adaptive shading. The display engine also has native support for HDMI 2.1 and DisplayPort 2.1, and the media engine has now been separated from the dead picture mode, adding support for things like AV1 encoding.
On the AI topic, Intel is also betting hard on Meteor Lake’s capabilities. All of the CPU, GPU, and NPU in Meteor Lake processors can be used for AI tasks, excelling in a variety of tasks. The GPU is better for AI tasks in media and 3D rendering, while the CPU is better for lighter AI tasks. NPU is dedicated to AI, providing the best performance in both standard AI work and offloading AI work.
The SoC and IO tiles are very nice and have what you would expect: the connection of important technologies such as USB, PCIe, Bluetooth, Wi-Fi, etc., but there are interesting points here. Meteor Lake will have Wi-Fi 7, Bluetooth 5.4, and Intel Unison technology for sharing the screen through the SoC tiles, which also has its own E-cores, although Intel did not say exactly how many. The IO tile has PCIe, Thunderbolt 4, USB, and other communication technologies. In addition, the SoC and IO tiles have the same functionality as AMD IO dies.
While it’s easy to call Intel chips the same as AMD chipsets, it’s important to note that Intel’s approach here is very different. AMD only generates IO dies, cache dies, and compute dies with CPU cores, and then uses multiple compute dies to increase throughput. Intel on the other hand doesn’t seem to be using multiple compute dies and CPU cores except maybe server chips, like Sapphire Rapids. Both approaches have pros and cons, and Intel is investing in its design thinking to be successful in the long run.
Intel’s strategy is moving forward: leadership by 2025
Intel’s 10nm node (now renamed to Intel 7) was an unmitigated disaster for the company, leading to several years of 14nm CPUs that slowly saw Intel lose its lead. Intel is now behind its biggest rival, TSMC, and Intel has a plan to catch up and return to first place. The first step is to get Intel 4 (which is what Meteor Lake’s Compute tiles are made of) out the door, which has twice the density of Intel 7 (formerly 10nm and the Raptor Lake process. Intel 4 will be followed by Intel 3). a straight version of Intel 4. These two nodes will be where Intel uses its best CPUs in 2024.
2025 will see the introduction of Intel’s 20A strategy, and this is what Intel believes will deliver success after a difficult decade for the company. The 20A will be followed by the 18A, and it looks like the 20A will be an upgrade in the same way that Intel 3 is better than the Intel 4. From previous statements, Intel will introduce desktop and laptop CPUs Arrow Lake and Lunar Lake on the 20A and 18A, which are planned for 2024 and later but it seems to be strong in 2025 based on the latest information from Intel.
Of course, one may remember that Intel’s attempt to make more progress in a short period of time is what caused the 10nm confusion in the first place, although this time Intel seems to be content to make progress in several areas instead of one node. , which would prevent what happened with 10nm. We’ll see if this helps Intel in the coming years.
Categories: Reviews
Source: thptvinhthang.edu.vn
