Intel’s 14th-gen cell processor, code-named Meteor Lake, represents a dramatic shift from the chips Intel has shipped for years — and Intel has supplied us with a ton of knowledge. If you don’t have time to wade via all of it, nonetheless, we’ve summed up crucial factors right here.
Here’s what you have to learn about Meteor Lake. For a way more detailed examination of what this new era of processors gives, please see our deep-dive into Meteor Lake.
1.) Big energy for half the facility
Intel historically has developed its processors alongside a “tick-tock” mannequin, the place an older chip was first migrated (“tick”) onto a brand new course of expertise, after which redesigned and sped up (“tock”). Michelle Johnston Holthaus, government vice chairman and normal supervisor of the Client Computing Group at Intel, was blunt: with Meteor Lake, Intel is focusing on the efficiency of its 13th-gen mobile Raptor Lake chips, however at half the facility (whoa). Any efficiency enhance shall be left to the graphics block, the place Intel is anticipating double the efficiency of the Raptor Lake built-in GPU with its new XeLPG built-in graphics core.
It reminds us of Qualcomm’s older messaging with its Snapdragon PC processors, really: related efficiency as an older Core chip, an enormous graphics enhance, and always-on connectivity. Intel is definitely providing each WiFi 7 and WiFi 6E in Meteor Lake, however with out the large advertising and marketing push Qualcomm has made.
2.) Intel builds Meteor Lake drastically totally different than previous CPUs
Traditional Intel “microprocessors” included each a CPU and a supplementary chip referred to as a PCH inside the identical bundle. Now, Intel is shifting to 4 chiplets or “tiles”: a CPU, “SOC,” graphics, and an I/O tile all inside the total Meteor Lake bundle. Intel calls this a “disaggregated” strategy, permitting it to plan every tile by itself growth roadmap and enhance the general yields.
But the graphics and SOC tiles are made by rival TSMC. Only the CPU and (possibly) the I/O tile are made by Intel, with the CPU tile particularly made on Intel’s Intel 4 process. Intel, feeling the sting of years caught on a 14nm node, has dedicated to quickly stepping via 5 course of applied sciences in 4 years.
3.) Meet Meteor Lake’s new low-power E-cores
With the 12th-gen Alder Lake, Intel moved from one sort of processor core to 2: beefy efficiency cores (P-cores) and smaller effectivity cores (E-cores) for background duties. Now there’s a 3rd sort of processor core. In addition to the brand new “Redwood Cove” P-core and the brand new “Crestmont” E-core, there’s a low-power model of the Crestmont E-cores that shall be used as the primary touchdown spot for duties which can be assigned to the CPU.
Intel isn’t saying how a lot sooner the P-cores are in comparison with these present in Raptor Lake. The Crestmont E-cores are between 4 to 6 p.c sooner, clock per clock. But right here’s one thing: Intel performed again a 4K video utilizing simply the low-power E-cores alone within the picture on the high of this story. That’s going to make a large distinction in your laptop computer’s battery life.
4.) No, severely, Meteor Lake is optimized for low energy
We’re not kidding. Intel’s Thread Director, the “butler” of the processor, takes duties that the working system assigns and routes them to the suitable cores. You might concentrate on how Thread Director worked in Raptor Lake: it could take duties or threads and push them to any efficiency cores (P-cores) that had been free. Raptor Lake pushed efficiency. In Meteor Lake, it’s precisely the alternative.
In Meteor Lake, new threads are routed to the low-power E-cores first, then the traditional E-cores, after which to the P-cores themselves. It’s a coverage change, and it’s an enormous one. Even in a “performance” state of affairs, new threads shall be routed to the traditional E-cores first, then the P-cores. Will it have an effect on the general efficiency of the chip? We don’t know.
5.) Welcome to the AI PC
The SOC tile is the place Intel’s NPU resides because the literal AI “brains” of the Meteor Lake chip. Intel’s NPU jogs my memory of the early days of GPUs: considerably fundamental, as each Intel, its clients, and builders determine what precisely AI is nice for. (Remember, that is native AI — Bing Chat, Google Bard and different AI engines reside within the cloud. Why even have native AI in your PC? Privacy and velocity, Intel says.)
The NPU, although, goes to energy what Intel calls the “AI PC.” Expect to listen to much more about that this week at Intel’s Innovation convention.
In AI, Intel’s actually simply making an attempt to speed up what it may: Windows capabilities, plus APIs just like the OpenVINO API that some builders use. Intel outlined what’s contained in the NPU at Innovation, however it gained’t make a lot sense — but — even for many fanatics who’re intimately acquainted with threads, cores, TDPs, cache hierarchies and the like. Over time, we’d anticipate Intel will determine what clients want and speed up it.
Mark Hachman / IDG
One fascinating level: accelerating AI with an NPU will be the most effective by way of energy consumption. But by way of uncooked velocity, a mixture of CPU, GPU, and NPU makes probably the most sense, Intel says.
6.) Meteor Lake’s graphics present a ray-tracing enhance
According to Intel, Meteor Lake takes the perfect elements of the XeHPG graphics expertise discovered inside Intel’s Arc graphics cards and places it inside Meteor Lake’s GPU. That means, briefly, a doubling of the built-in graphics efficiency from Raptor Lake to Meteor Lake, in addition to built-in ray tracing assist.
While Intel isn’t releasing benchmarks but (and we don’t have the chip obtainable for testing), Intel shared a few of its “micros,” or inside benchmarks: triangle draw charge improves 2.6X over Raptor Lake, with pixel mix charges enhancing by 2.1X. While evaluating the depth check charge, nonetheless, Intel noticed a 6.6X enchancment.
The Meteor Lake GPU additionally helps XeSS, Intel’s reply to the AI-powered DLSS expertise that upsamples a lower-resolution picture to enhance picture processing. And there’s one thing new by way of low-power (there’s that phrase once more!) gaming, too: Endurance Gaming, which is able to enable avid gamers on the go to increase their laptop computer’s battery life dramatically. According to Intel execs, the Meteor Lake GPU can render Rocket League at lower than 1W of energy.
What are Meteor Lake’s specs?
They’re not right here but, a minimum of formally. Intel sometimes explains how its processors work first, then gives all of the essential shopping for data you’ll want, later. Unfortunately, you’ll have to attend till nearer to the official launch (sometimes on the CES present in Las Vegas in January) for the official speeds and feeds.
We do know that Intel confirmed off a six P-core, eight E-core implementation of Meteor Lake in Malaysia. Unfortunately, Intel is saying solely that it was for “demonstration purposes” and extra data shall be shared nearer to launch.
In the meantime, we’ll look ahead to AMD’s Ryzen 8000 collection chips to roll out subsequent yr, and information of Qualcomm’s Oryon collection (based mostly upon its Nuvia processor expertise) to hopefully be disclosed in October.
(Disclosure: whereas we anticipated Intel to launch Meteor Lake at its Intel Innovation convention, it as a substitute flew journalists and analysts to its meeting and check services in Penang, Malaysia. Intel paid for PCWorld to attend, together with airfare, resort and meals. To acquire entry to this data, PCWorld accepted, although it maintained editorial independence all through and doesn’t have an effect on our Meteor Lake protection.)