Retail Boxes of Intel Core Ultra 200-series "Arrow Lake" 65W Processors Surface

[btarunr]

Here are some of the first pictures of the retail boxes of the Intel Core Ultra 200-series "Arrow Lake-S" 65 W desktop processors. Intel debuted the series with Unlocked K-series SKUs in October 2024, and will expand it in January 2025 with 65 W models that lack unlocked multipliers. The unlocked models lack any included stock cooling solution, while the 65 W models come with them. There's no word on what the coolers look like, but if we were to guess, Intel will reuse its Laminar series fan-heatsinks that it debuted with its 12th Gen Core "Alder Lake."

The new Socket LGA1851 retains cooler compatibility with the previous LGA1700, which is why the company could reuse the Laminar series. The Core Ultra 9 65 W retail box appears the largest, and so it could include the Laminar RH1 cooler that's capable of cooling the processor as it draws its maximum turbo power. This cooler comes with some blue LED illumination. The Core Ultra 7 and Core Ultra 5 65 W retail boxes appear to have the same thickness, which means the two could include the mid-tier Laminar RM1 cooler. The RM1 has a slightly thinner heatsink, but comes with an illuminated ring along the bore of the fan-frame. If Intel launches a Core Ultra 3 series (successor to the Core i3), Intel will likely include the Laminar RS1, the lightest variant, which lacks any lighting. The first three pictures (below) show boxes of the standard variants of the 65 W SKUs, the 4th and 5th pics show the boxes of the "F" SKUs which lack integrated graphics.



View at TechPowerUp Main Site | Source

 

[yfn_ratchet]

I thought the naming convention was going to delienate between performance segments, i.e. higher wattage/boost clock parts would be the "Core Ultra ___ (K/KF)" and lower end parts would just be "Core ___(F)". What purpose does the 'Ultra' serve if it says nothing of what the product's target audience would be? "Core [Architecture] ___(K/KF/F)" would have been better. "Core Arrow 285K" says a lot more about the product than just a generic "Ultra" moniker.

 

[kondamin]

Shame they are bothering with the entire lineup, it’s not going to be a successful series.

 

[yfn_ratchet]

Shame they are bothering with the entire lineup, it’s not going to be a successful series.

It's the expense of the platform compared to the performance of the chips, IMO. If Arrow Lake came out swinging with ~$110 B-series boards and ~$160 240F chips it would have been a great early-adopter package for whatever they have cooking for the socket later down the line (so obviously Intel would have to commit to the socket for at least a gen and a half for the sell to work). Passable punch at passable wattages, at a passable price. Not a game-changer, but hey, you don't have to wait years to buy in cheap like with AM5.

 

[Chaitanya]

Shame they are bothering with the entire lineup, it’s not going to be a successful series.

They have already sunk the money in the series also these lower end chips are the bread and butter for OEM and SIs.

 

[TheDeeGee]

Shame they are bothering with the entire lineup, it’s not going to be a successful series.

I went with a 11700 non-K thinking it's fine for gaming, which it is to an extend. But when you also try to do some basic streaming on the side even though it's on the NVENC encoder it tends to not be enough.

I currently have it unlocked to 110W (not ideal for non-K) instead which fixes it all, and it's stable. But my next upgrade will be a K model again which is designed for higher wattage.

Sometimes you just got to find things out for yourself in life you know ^^

 

[Vincero]

I went with a 11700 non-K thinking it's fine for gaming, which it is to an extend. But when you also try to do some basic streaming on the side even though it's on the NVENC encoder it tends to not be enough.

I currently have it unlocked to 110W (not ideal for non-K) instead which fixes it all, and it's stable. But my next upgrade will be a K model again which is designed for higher wattage.

Sometimes you just got to find things out for yourself in life you know ^^

This used to work fine 10 years ago (e.g. if you had say an i7-3770k/non-K and weren't overclocking there was barely any difference between them) but as Intel have now reached some hard process limits and need to blast past 65/95W TDPs to achieve a high all-core performance it is more of a differentiator at the high end. At the low end it's less of an issue.

The was the problem with getting closer to the limits of the silicon and Coffee/Rocket Lake was definitely pushing the limits - the all-core speeds were so low whilst maintaining the TDP window and the turbo speeds would only be maintained with way over base level TDP ratings - due to using such an old process it really couldn't maintain that performance level.

Even with Intel's current Raptor Lake parts, they can only do their max performance tearing past 200W TDP. Ironically, if they run at half that they lose only a minor amount of performance.

If you take the 11700 K Vs non-K, running both at their spec settings, the benefit of the K having 125W TDP means it can hit the rated speed of 3.6GHz on all cores Vs 2.5GHz for the 65W part.
Other than that the boost speeds are only 100MHz different between them, however the K will likely hold those a lot better.
For low thread workloads there isn't that much between them, but add more workload and it's the K that should come out on top, so long as it's cooling is sufficient.

Of course the K part can be manually overclocked but most motherboards mess around (and do not follow / exploit Intel guidance) with the PL1/2 and power/current limits so that even non-K parts can achieve and maintain all core boost speeds in excess of their normal time rating (usually so long as it remains under some temperature and power limits) - this is something I've seen going all the way back to Sandy Bridge - on a good board with good cooling (definitely required with newer Intel parts even with their improved laminar box cooler, i.e. it can still hold you back) this was common until now with the Raptor Lake issues and Intel basically throwing motherboard vendors under the bus (whilst at the same time not penalising them for doing this for years now as it obviously made Intel solutions look better).

Intel should really not number the non-K and K parts the same anymore. The equivalency between them is long gone. Whereas the K used to denote mostly 'hey you can mess around with it', that really isn't the case today.

 

[Vayra86]

No way, the box is BLUE?!

 

[Palladium]

The was the problem with getting closer to the limits of the silicon and Coffee/Rocket Lake was definitely pushing the limits - the all-core speeds were so low whilst maintaining the TDP window and the turbo speeds would only be maintained with way over base level TDP ratings - due to using such an old process it really couldn't maintain that performance level.

I live in a tropical climate, my old 8700K couldn't even run P95 at 4.3GHz all-core stock on a Deepcool Gammaxx without throttling.

4790K was my last CPU that actually run P95 at full turbo clocks out of the box with a mainstream 4-pipe 120mm cooler.

 

[kondamin]

They have already sunk the money in the series also these lower end chips are the bread and butter for OEM and SIs.

a good chunk of revenue goes to tsmc this round redirecting capacity to gpu's instead of 26 sku's from which 20 they make maybe a dollar in profit

 

[wNotyarD]

I thought the naming convention was going to delienate between performance segments, i.e. higher wattage/boost clock parts would be the "Core Ultra ___ (K/KF)" and lower end parts would just be "Core ___(F)". What purpose does the 'Ultra' serve if it says nothing of what the product's target audience would be? "Core [Architecture] ___(K/KF/F)" would have been better. "Core Arrow 285K" says a lot more about the product than just a generic "Ultra" moniker.

For Intel, Ultra means AI (NPU inside). Hence, Meteor, Lunar and Arrow are Ultras.

 

[Wirko]

a good chunk of revenue goes to tsmc this round redirecting capacity to gpu's instead of 26 sku's from which 20 they make maybe a dollar in profit

Any chance they're going to make lower-end Arrow processors themselves? Intel did mention the possibility of dual-sourcing a few times.

 

[Drash]

Anyone else totally bewildered by this xyz-lake maybe S or F or K or ??? naming (ok I get K but ...). I read articles that say xxx lake, with yyy arch and then mention zzz generation, and think WTF how do I position them in a timeline. Which gen is which lake? Which lake uses what arch? Are we talking about stuff already out, coming out soon, further future stuff that might get canned anyhow? Why do I need to know which gen which lake is. Is a new lake a new gen? Is a new arch a new lake and/or a new gen? Bring back incrementing number + K.

Looks like my trusty 6700K is getting replaced by an AMD X3D, when they get round to making some that I can buy, partly because it's better for gaming, and partly because I don't get a headache every time I think about the name.

 

[kondamin]

Any chance they're going to make lower-end Arrow processors themselves? Intel did mention the possibility of dual-sourcing a few times.

The entire architecture needs a make over, the IO die and memory controller need to be next to each other not at opposite ends and and that's just the most obvious issue.
Even if they managed to get 18nm going, it would be far more interesting they moved on to the next architecture with that and treat arrow lake the same they did rocket lake.

 

[dartuil]

Are they cheap now?
In my country that socket cost your leg