News Intel announces Arrow Lake and Lunar Lake will arrive in 2024 with 3 times more GPU and AI acceleration performance

[Admin]

Intel revealed that its Arrow Lake and Lunar Lake processors will come to market in the second half of 2024.

Intel announces Arrow Lake and Lunar Lake will arrive in 2024 with 3 times more GPU and AI acceleration performance : Read more

 

[Findecanor]

Serious question: Why in the world would I want that?

 

[cyrusfox]

For GPU power, encode, decode and gaming, otherwise yes you don't need it

Serious question: Why in the world would I want that?

For AI- Windows 12 AI features
Now personally I like the custom accelerators now available to all that will speed up creator task (media editing/creation). The window features... garbage honestly. Microsoft continues to break its own search since Windows 7, then to fix it again in windows 10/11.

 

[circadia]

I hope the NPU can be used for other purposes, especially for something like AI frame-gen.

 

[usertests]

Hard to tell without more numbers but sounds like they're trying to get to the same 45-50 TOPS ballpark as AMD's Strix Point and Qualcomm's Snapdragon X Elite.

This chip is largely thought to be an unreleased Meteor Lake revision with on-package LPDDR5X, much like the Lunar Lake model Intel displayed today.

I assume the unreleased Core Ultra 7 164U and Core Ultra 5 134U will be the models with on-package LPDDR5X. Those have a lower 9W base TDP.

 

[Geef]

Make sure your PC AI isn't listening to you when you talk. You don't want to be playing a video game with your friend yelling at him goofey phrases like your gonna beat him with a rainbow di_do if he kills you again and later have Amazon show you the availability of that item on the front page!

 

[DSzymborski]

Serious question: Why in the world would I want that?

I don't need a vehicle that can tow a boat, but I don't go around feigning ignorance about why there are vehicles that have the ability to tow a boat.

 

[peachpuff]

Intel should use ai to increase their ipc...

 

[spicy_cat]

Hard to tell without more numbers but sounds like they're trying to get to the same 45-50 TOPS ballpark as AMD's Strix Point and Qualcomm's Snapdragon X Elite.


I assume the unreleased Core Ultra 7 164U and Core Ultra 5 134U will be the models with on-package LPDDR5X. Those have a lower 9W base TDP.

Do we know if the on-package LPDDR5X will be used in place of main RAM for tablet/ultra-thin laptop form factors or if it's meant to be an extra-wide memory bus close to the CPU as an extra level of cache to compete with X3D? Or can it do both duties like the embedded DRAM in some of Intel's server CPUs?

 

[magbarn]

Same 6 Pcore limit? No Thanks. Who cares about AI. They need to release a CPU that's at least competitive with the 7XXXX3D chips in gaming and desktop meteor is not it.

 

[vehekos]

I hope the NPU can be used for other purposes, especially for something like AI frame-gen.

It surely will serve to run AI in the user PC. local chatgtp, copilot.

 

[emike09]

I don't need a vehicle that can tow a boat, but I don't go around feigning ignorance about why there are vehicles that have the ability to tow a boat.

I respect analogies, but that's a terrible one in this case.

These are mainstream processors. I also beg the question - Why do we need AI hardware in a CPU? What will they bring the average consumer? Especially with MS trying to move computing to the cloud.

I could see local AI being useful somehow, but I fail to understand how. It all just feels like marketing and fluff. Nvidia has made it useful to the gaming world with DLSS, FrameGen, etc, but how will Intel do the same?

 

[DSzymborski]

I respect analogies, but that's a terrible one in this case.

These are mainstream processors. I also beg the question - Why do we need AI hardware in a CPU? What will they bring the average consumer? Especially with MS trying to move computing to the cloud.

I could see local AI being useful somehow, but I fail to understand how. It all just feels like marketing and fluff. Nvidia has made it useful to the gaming world with DLSS, FrameGen, etc, but how will Intel do the same?

Which would make sense if something else was being taken away. People use CPUs for a vast array of things, and generalized CPUs need to be as good at the vast array of things as possible. CPUs aren't buffets.

 

[usertests]

Do we know if the on-package LPDDR5X will be used in place of main RAM for tablet/ultra-thin laptop form factors or if it's meant to be an extra-wide memory bus close to the CPU as an extra level of cache to compete with X3D? Or can it do both duties like the embedded DRAM in some of Intel's server CPUs?

I don't think it acts any different from main memory, and all of it will most likely be on the package. It may cut costs, increase achievable LPDDR5X speeds, and lower power consumption, all slightly.

I think I know the Xeon products you're referring to, where you can use the HBM without external DDR5 memory, or as a cache.

 

[ThomasKinsley]

If Lunar Lake will have 3 times the AI processing power of Meteor Lake and the chips are already being shipped to Intel's partners, then what is the reason to buy Meteor Lake? The NPU was the only unique aspect and it's being superseded quickly.

 

[aero1x]

Is this the -back to you Steve lady? Back to you Steve!

 

[usertests]

Same 6 Pcore limit? No Thanks. Who cares about AI. They need to release a CPU that's at least competitive with the 7XXXX3D chips in gaming and desktop meteor is not it.

Desktop Arrow Lake should have 8 P-cores, not sure where you're getting that:
https://videocardz.com/newz/intel-r...-refresh-featuring-8p32e-cores-for-2025-debut

 

[TerryLaze]

I respect analogies, but that's a terrible one in this case.

These are mainstream processors. I also beg the question - Why do we need AI hardware in a CPU? What will they bring the average consumer? Especially with MS trying to move computing to the cloud.

I could see local AI being useful somehow, but I fail to understand how. It all just feels like marketing and fluff. Nvidia has made it useful to the gaming world with DLSS, FrameGen, etc, but how will Intel do the same?

Already part of DirectX so it will be used by devs for all kinds of stuff, we already have big software suits like adobe and such that do a lot of stuff on (i)GPUs and this will be just like that, if you have it it will accelerate a lot of things and if you don't have it you will be forced to wait for hours for results.
Also webcam background removal, picture sorting and many video and picture filters can use AI, as can audio software, there are plenty of things, also with these chips finally being good enough for gaming a handheld console based on them could use it for DLSS, framegen, etc.

DirectML: Accelerating AI on Windows, now with NPUs - DirectX Developer Blog

We are thrilled to announce our collaboration with Intel®, one of our key partners, to bring the first Neural Processing Unit (NPU) powered by DirectML on Windows. AI is transforming the world, driving innovation, and creating value across industries. NPUs are critical components in enabling...

devblogs.microsoft.com

 

[Findecanor]

I don't need a vehicle that can tow a boat, but I don't go around feigning ignorance about why there are vehicles that have the ability to tow a boat.

Sure, but so far we have hardly seen any boats connected to PCs. We have seen telepresence connections to larger boats elsewhere.

What smaller, local, boats do they expect there to be? Who is making them? Where can I read about them?

 

[bit_user]

If Lunar Lake will have 3 times the AI processing power of Meteor Lake and the chips are already being shipped to Intel's partners, then what is the reason to buy Meteor Lake? The NPU was the only unique aspect and it's being superseded quickly.

Buy Meteor Lake, if you can't afford to wait and want an Intel laptop that's more power-efficient or has a better iGPU than the laptops featuring their Gen12 & Gen 13 CPUs.

However, if you can afford to wait, there's always something better that's just around the corner. That's just how it is, with tech.

 

[DSzymborski]

Sure, but so far we have hardly seen any boats connected to PCs. We have seen telepresence connections to larger boats elsewhere.

What smaller, local, boats do they expect there to be? Who is making them? Where can I read about them?

This has nothing to do with boats and you know it.

 

[kwohlt]

Serious question: Why in the world would I want that?

I don't know why you personally would want one or what your workloads look like, but these CPUs will be sold in the 10s of millions, and NPUs will benefit many of us.

Use cases for local NPUs have been talked about a lot. It includes Blurred backgrounds in video calls, foreground/background cropping, text-to-speech and speech-to-text, local translation, OCR, video and photo editing - these are fairly common tasks many people do routinely, and perf/watt increases quite a bit when using an NPU.

And less common tasks could include everything from those face swap apps, to CoPilot, to Stable Diffusion, or ChatGPT. Use cases for NPUs will only continue to grow as people actually start having them.

 

[kwohlt]

If Lunar Lake will have 3 times the AI processing power of Meteor Lake and the chips are already being shipped to Intel's partners, then what is the reason to buy Meteor Lake? The NPU was the only unique aspect and it's being superseded quickly.

Because most Lunar Lake laptops aren't going to be available until next year's CES.

although I'm sure a small handful of models might release next Nov. / Dec.

 

[kwohlt]

I respect analogies, but that's a terrible one in this case.

These are mainstream processors. I also beg the question - Why do we need AI hardware in a CPU? What will they bring the average consumer? Especially with MS trying to move computing to the cloud.

I could see local AI being useful somehow, but I fail to understand how. It all just feels like marketing and fluff. Nvidia has made it useful to the gaming world with DLSS, FrameGen, etc, but how will Intel do the same?

If "needs" were the only thing we were concerned about, thin-and-light CPU/GPU performance could have remained completely unchanged for the last 6 years, and we'd meet the needs of 80%+ of the market.

A need doesn't need to be first identified before next-gen CPUs improve perf by adding accelerators that'll improve the workloads of many people just because it may not personally benefit yours.

 

[Greg7579]

I'm confused. I want to upgrade my laptop to the Dell XPS 16 with Intel Core ultra 9 185H (Meteor Lake) coming out in February, but now a couple of weeks later they announce that Lunar Lake is coming to laptops this year! Why would anyone buy Meteor Lake, which is coming out right now on high-end (and other) laptops when Lunar Lake will have 3 times the AI processing power of Meteor Lake?
Well, I just built a top-end PC a month ago using the refresh chip from Gen 13 (i9 14900K) when I know a generational change (Arrow Lake) for the PC is probably 8 or 9 months away. Same thing, I guess. But that Meteor Lake to Lunar Lake window is very tight. Why get Meteor Lake now?