Intel Arrow Lake-S desktop CPU leak includes a 24-core 285K flagship, 20-core 265K

[DragonSlayer101]

Rumor mill: A recent report claimed that Intel's next-gen Arrow Lake-S desktop processors could be marketed under the "Core Ultra 200" branding. These chips are said to have been designed for the so-called AI PCs that Microsoft believes could bring a tectonic shift in the world of personal computing. A new leak now claims to have revealed the specifications of some of these CPUs.

According to Chinese tech publication Benchlife, Arrow Lake-S will comprise 13 SKUs at launch, including three K-series models designed for overclocking. The first of these is the Core Ultra 9 285K, which is said to be a 24-core chip with an 8+16 configuration, while the other two are the Core Ultra 7 265K with 20 cores (8+12) and Core Ultra 5 245K with 14 cores (6+8). All of them will reportedly have a 125W TDP.

The lineup is tipped to further include three non-K parts with locked multipliers. An earlier report from leaker @OneRaichu mentions them as "275," "255," and "240," but we can expect them to be marketed as the Core Ultra 9 275, Core Ultra 7 255 and Core Ultra 5 240. All of them will be 65W parts and at least one of them – the Core Ultra 5 SKU – is rumored to have a 10-core design with a 6+4 configuration. Unfortunately, none will offer hyper-threading, as Intel is deprecating the feature with the Arrow Lake-S processors.

The lineup is also expected to include two Core Ultra 200F-series chips. One will have a 125W TDP, and another will feature a 65W power design. Additionally, Arrow Lake-S will offer five SKUs as part of the Core Ultra 200T-series, all tipped to be 35W parts. The report does not include any other information about the upcoming processors, such as clock speeds, power limits, and memory speed support, but we'll likely discover those details in the lead-up to their launch.

According to earlier reports, the Core Ultra 200 series processors will only be compatible with new 800-series motherboards featuring the LGA-1851 socket, meaning you'll have to spend some additional money to update to the new platform. Recent leaks suggest that Intel could announce the Arrow Lake-S lineup at the Computex 2024 event in Taiwan next month, so we may not have to wait too long to get our hands on the official spec sheets.

Permalink to story:

Intel Arrow Lake-S desktop CPU leak includes a 24-core 285K flagship, 20-core 265K

 

[Theinsanegamer]

STILL no more advancement in core config. 8p is the new quad core.

So, any guesses as to actual power use on the i9? My guess is up to 320 watts.

 

[Kn0xx]

Does Intel live under the rock, and don't see that the competition is getting far ahead ?
why Intel offerings are un-impressive ?

 

[LetTheWookieWin]

I’ll reserve judgement until we see the benchmarks and power/thermal performance.

I’m very interested to see how the removal of hyperthreading is sold. Presumably they needed die space for something else. That or it’s a marketing ploy and they’ll reintroduce it later.

 

[PandaBacon]

What a mess in the naming. Why not have the 9s be the 29x, 7 27x, 5 25x… it’s gonna be interesting how reviewers are gonna be calling the cpus, after hearing 14-900-k for so long. 2-85k?

 

[human7]

I’ll reserve judgement until we see the benchmarks and power/thermal performance.

I’m very interested to see how the removal of hyperthreading is sold. Presumably they needed die space for something else. That or it’s a marketing ploy and they’ll reintroduce it later.

Die space for their NPU, perhaps.

In fairness, hyperthreading isn't really needed when you have ample core quantities to begin with, and I suppose cache closer to the core could be put there instead. We will see. If the benchmarks are good and AMD also responds (this year or next) by cutting SMT, then I would say there is merit to the new architecture.

 

[daffy duck]

STILL no more advancement in core config. 8p is the new quad core.

So, any guesses as to actual power use on the i9? My guess is up to 320 watts.

Come on Arrow Lake will use a lot less power than cRaptor Lake. Ludicrous to think it would use more. Power efficiency is the whole reason Intel is using tiled structures and moving to PowerVIA. An I say this as someone that does not currently use Intel and thinks there current chips are a bit of a joke.

 

[LetTheWookieWin]

Die space for their NPU, perhaps.

In fairness, hyperthreading isn't really needed when you have ample core quantities to begin with, and I suppose cache closer to the core could be put there instead. We will see. If the benchmarks are good and AMD also responds (this year or next) by cutting SMT, then I would say there is merit to the new architecture.

Yeah, the NPU was what came to my mind as well. I’m hoping the NPU is useful, but I can’t help but wonder if it’s going to be like the iGPU (I.e., a waste of die space if you already own a good GPU).

I agree that it will be interesting to see whether AMD makes similar design decisions.

 

[Rudyduu]

Come on Arrow Lake will use a lot less power than cRaptor Lake. Ludicrous to think it would use more. Power efficiency is the whole reason Intel is using tiled structures and moving to PowerVIA. An I say this as someone that does not currently use Intel and thinks there current chips are a bit of a joke.

How can you make assumptions with no data? None of these brands care about you, so why does it matter what brand the CPU is? Buy the best CPU you can justify spending money on and stop caring about the brand. It's "their" not "there" in this case.

 

[Rudyduu]

Yeah, the NPU was what came to my mind as well. I’m hoping the NPU is useful, but I can’t help but wonder if it’s going to be like the iGPU (I.e., a waste of die space if you already own a good GPU).

I agree that it will be interesting to see whether AMD makes similar design decisions.

The iGPU accelerates video encoding and is a great troubleshooting tool for identifying GPU issues.

 

[HardReset]

Die space for their NPU, perhaps.

In fairness, hyperthreading isn't really needed when you have ample core quantities to begin with, and I suppose cache closer to the core could be put there instead. We will see. If the benchmarks are good and AMD also responds (this year or next) by cutting SMT, then I would say there is merit to the new architecture.

Hyper Threading directly increases die area about 5-10 percent. leaving it out just because of that makes no sense. It probably won't have huge impact on cache latency either.

About only reason that makes any sense is Intel's idea to offload background processes on Crap cores. That way Hyper Threading does ot lower performance on low core usage situations like on games since Intel uses no priority on HT (thread that goes on HT "core" and real core is basically random).

Problem is, putting background processes on Crap cores is stupid fron beginning. It seems Intel is continuing stupidity after they decided to go that way.

For AMD, their SMT is more effective than Intel's, less reasons to get rid of it.

 

[LetTheWookieWin]

The iGPU accelerates video encoding and is a great troubleshooting tool for identifying GPU issues.

Agreed…. but if I had to choose between HT and the iGPU on my current CPU I’d pick the former in a heartbeat.

 

[bennyg]

Hmmm I smell a rat. Intel's new lower power 'crash safe' settings packages - and the loss of up to 10% performance - are going to have an additional bonus of boosting ARL's "generational improvements" marketing.
Take the hit on reputation with a nerf to products already sold, and reduce RMA rates by offloading responsibility to mobo vendors... to boost the marketing of the next gen (ARL will need all the help it can get against the zen5 wall it's going to hit)