News Intel Bartlett Lake-S: up to 12P-Core or up to 8P-Core +16E-core

[jpiniero]

but even a full 12P with AVX512 will be utterly crushed by AM5 EPYC 4005, albeit it is significantly less worse than current offerings

That is very true.

 

[igor_kavinski]

That was not the question. The question was if any of THIS list of Processors (Or any E for that matter) works on consumer motherboards.

Again those CPUs would be restricted based on BIOS code only. These Core E CPUs would only be supported in the BIOS of industrial/embedded mobos like this: https://www.asus.com/networking-iot...80ea-im-a/helpdesk_cpu?model2Name=R680EA-IM-A

This DFI (yes, the LANparty company!) mobo mentions Bartlett Lake support: https://www.dfi.com/product/index/1612#specification

It's all BIOS restricted it seems.

 

[zir_blazer]

Again those CPUs would be restricted based on BIOS code only. These Core E CPUs would only be supported in the BIOS of industrial/embedded mobos like this: https://www.asus.com/networking-iot...80ea-im-a/helpdesk_cpu?model2Name=R680EA-IM-A

This DFI (yes, the LANparty company!) mobo mentions Bartlett Lake support: https://www.dfi.com/product/index/1612#specification

It's all BIOS restricted it seems.

This is still not the answer I'm looking for.
There is a gaping hole between "officially supported", "not officially supported, but works" and "intentionally blacklisted to not work". We do NOT know if it is BIOS restricted or not because there is NO evidence of anybody actually TRYING one of those in your average Z690/Z790 consumer board to report that it doesn't POST. This is important because the currently released Bartlett Lake-S SKUs are for IoT like the board you posted, and I don't recall any leak that specifically mentioned that there will be non-IoT SKUs (Including the 12P die everyone is waiting for). So we have no proof of whenever the E/IoT models are BIOS restricted or not because there is no record of anyone trying to POST with those on a consumer board, which may become reelevant if Intel decides to not make a consumer launch.

 

[igor_kavinski]

We do NOT know if it is BIOS restricted or not because there is NO evidence of anybody actually TRYING one of those in your average Z690/Z790 consumer board to report that it doesn't POST.

TBH, the E CPUs don't have anything interesting about them and they are also not easily available. Worse, they are expensive when you do find them on some obscure shopping site, mostly from Europe. So not surprised that no one has gotten their hands on them to try one out in a consumer mobo.

So we have no proof of whenever the E/IoT models are BIOS restricted or not because there is no record of anyone trying to POST with those on a consumer board, which may become reelevant if Intel decides to not make a consumer launch.

Obviously we won't be able to use the 12P CPU if Intel forbids us to use it in a consumer mobo. Only time will tell I guess.

I think one reason Intel could be keeping the 12P CPU hush hush is because they really, really want to exhaust their Arrow Lake inventory.

 

[LightningZ71]

If they really wanted to have a leading gaming part, they have the needed building blocks all in place. They could have done Bartlett Lake on Intel 3, using 12 Redwood Cove cores with 4 MB L3 per core. To really get there, they could have done two of the cores with increased buffer space for extra clock range.

 

[igor_kavinski]

They could have done Bartlett Lake on Intel 3

I think this is the dirty secret that Pat and Intel don't want the world to know. Intel 4 and Intel 3 are very expensive compared to Intel 7 and not feasible for volume production. I think it's the same thing with Intel 18A. These technologies have become so expensive that Intel cannot hope to get back their development costs on them. That's why they are still going to try one more time to milk Intel 7 with Bartlett Lake until the yields and capacity for 18A improve to their satisfaction.

 

[DrMrLordX]

I think this is the dirty secret that Pat and Intel don't want the world to know. Intel 4 and Intel 3 are very expensive compared to Intel 7 and not feasible for volume production. I think it's the same thing with Intel 18A. These technologies have become so expensive that Intel cannot hope to get back their development costs on them. That's why they are still going to try one more time to milk Intel 7 with Bartlett Lake until the yields and capacity for 18A improve to their satisfaction.

Bartlett Lake is probably not going to ship in significant volume. It's likely to be a niche product. Also depending on whom you ask, some have indicated that Intel 7 is still pretty expensive and that Intel 3 might actually be cheaper per transistor.

 

[LightningZ71]

The party line is that Intel's capacity to produce product on Intel3 is very limited and booked with internal products.

 

[AcrosTinus]

I think this is the dirty secret that Pat and Intel don't want the world to know. Intel 4 and Intel 3 are very expensive compared to Intel 7 and not feasible for volume production. I think it's the same thing with Intel 18A. These technologies have become so expensive that Intel cannot hope to get back their development costs on them. That's why they are still going to try one more time to milk Intel 7 with Bartlett Lake until the yields and capacity for 18A improve to their satisfaction.

I think you are right and wrong at the same time. All nodes are expensive and the scale is what brings them down in cost. The big Idea behind the fab business is to not be the only customers and have everyone interested carry a piece of the burden. Look at TSMC, their nodes are super expensive as well, but you have a Nvidia that gives them a call and basically says "for x product, we might need y specs, can we customize node z for that purpose we are willing to invest XX amount into the development", the same goes on with Apple and every big customer. If everyone's future products success depends on node YY working out, they will make it happen, due to the resources being pooled and the risk shared.

With Intel they carry the burden and the associated risks and we saw and are still seeing how it worked out and the long tail effects. In order to survive with a fab Intel needs almost majority market share, share the fab with others (their current plan) or slowly start derisking by pushing essential products to foreign nodes (TSMC) which they are doing as well. So Intel has a plan, it is being executed but it is almost like a rotten arm that the patient does not want to cut off because the doc says its 50:50 for full recovery.

In short panther needs to scale for 18A to make sense and the rumored 12P core CPU on LGA1700 might never come out with Lip Bu at the head, the way I read him, he might be a penny pincher basically cutting down the company to a more agile core.

 

[igor_kavinski]

the rumored 12P core CPU on LGA1700 might never come out with Lip Bu at the head, the way I read him, he might be a penny pincher basically cutting down the company to a more agile core.

At the same time, if their plan is to entice or lure existing LGA1700 gamers towards the 12P SKU as a decent gaming performance upgrade without platform change, it could work. Most gamers still rocking 12900K/13900K/14900K wouldn't have an issue ditching their existing CPU for the 12P one since they don't care about the E-cores anyway. And a lot of these rabid gamers are very allergic to the X3D CPUs, with some of them switching to them and back to their Alders/Raptors because they felt something was "off".

 

[poke01]

And a lot of these rabid gamers are very allergic to the X3D CPUs, with some of them switching to them and back to their Alders/Raptors because they felt something was "off".

Funny cause next gen only AMD will have a 12 P core X3D CPU on a single die that’s actually worth it. Wonder what the excuse from these “gamers” will be then.

 

[eek2121]

At the same time, if their plan is to entice or lure existing LGA1700 gamers towards the 12P SKU as a decent gaming performance upgrade without platform change, it could work. Most gamers still rocking 12900K/13900K/14900K wouldn't have an issue ditching their existing CPU for the 12P one since they don't care about the E-cores anyway. And a lot of these rabid gamers are very allergic to the X3D CPUs, with some of them switching to them and back to their Alders/Raptors because they felt something was "off".

My previous understanding was that these chips aren’t DIY focused at all. They are targeting certain embedded, OEM, and industrial markets. Also some low end - low midrange type stuff.

I am behind on this one, however, so my info may be out of date.

 

[igor_kavinski]

I am behind on this one, however, so my info may be out of date.

Our hope is hanging by this thin thread: https://forums.anandtech.com/thread...-up-to-8p-core-16e-core.2620676/post-41469169

Apparently, matches the specs of Xeon E but with lower clocks.

 

[Wolverine2349]

The point of doing the 12 core die is primarily for Xeon E.

We'll see if they re-enable AVX-512.

Is Xeon E for entry level servers?

In such case seems unlikely there will be an unlocked version as Xeon E CPUs are locked. Hope to be

Funny cause next gen only AMD will have a 12 P core X3D CPU on a single die that’s actually worth it. Wonder what the excuse from these “gamers” will be then.

I feel something seems off with responsiveness and latency with AMD 9800X3D.

I went back to 14700K as such. Feels smoother and snappier in games even at 4K anyways with a RTX 5090.

 

[igor_kavinski]

I feel something seems off with responsiveness and latency with AMD 9800X3D.

I went back to 14700K as such. Feels smoother and snappier in games even at 4K anyways with a RTX 5090.

Maybe that's tile latency you are sensitive to? If so, then you need to buy a few backup 14700K CPUs because that's the last monolithic CPU Intel will ever produce, other than maybe cheap monts.

You could also try the upcoming Gorgon Point 9000G APU to see if "feels" better than an X3D.

 

[poke01]

Maybe that's tile latency you are sensitive to? If so, then you need to buy a few backup 14700K CPUs because that's the last monolithic CPU Intel will ever produce, other than maybe cheap monts.

Lunar lake shows Intel can do better. Arrow lake is held back due to bad design.

feel something seems off with responsiveness and latency with AMD 9800X3D.

Zen 6 should improve chiplet latency

 

[igor_kavinski]

In daily use, I couldn't find much difference in the snappiness of 9950X3D and 245KF. Both felt the same. Just fast. Very fast.

 

[Thunder 57]

Is Xeon E for entry level servers?

In such case seems unlikely there will be an unlocked version as Xeon E CPUs are locked. Hope to be


I feel something seems off with responsiveness and latency with AMD 9800X3D.

I went back to 14700K as such. Feels smoother and snappier in games even at 4K anyways with a RTX 5090.

At 4k X3D does a whole bunch of nothing in most games. Did you do any benchmarking? Snappiness is quite subjective is not a term I've typically heard regarding games.

 

[dullard]

Snappiness is quite subjective is not a term I've typically heard regarding games.

You haven't followed the trend towards small, ultralightweight mice on the smoothest surface possible, with ultra-high laser resolution, etc.

For example: "So for shooters, action games, and things that require an ultra snappy response, I'll go light for sure."

Five things I always tell people before they buy their first gaming mouse

My guide to buying the right gaming mouse for you.

www.pcgamer.com

Same with high refresh rate monitors. Or monitors with a short refresh time. Or high end video cards. Many games are all about snappiness. Many gamers feel that even one millisecond of lag will cause them to miss the target--whether that is actually true or not.

Yes, snappiness is not easy to measure quantitatively. But Snappiness is the whole reason for P cores. Get the User Interface tasks done ASAP so things feel fast and responsive.

 

[zir_blazer]

Yes, snappiness is not easy to measure quantitatively. But Snappiness is the whole reason for P cores. Get the User Interface tasks done ASAP so things feel fast and responsive.

To me this was always a Software problem. Wait until gamers learn about Real Time Operating Systems like QNX or a very well tuned Linux with the RT patches, Core isolation so that they perform one specific thread/application with no CPU Scheduler shenanigans, along with an objetive way to measure latency (Perhaps someone still remember some extremely complicated raw recording setup with RAID 0 of SSDs that reviewers used to get the raw video output from a pair of Radeon HD 5870 in CrossFire to prove that CrossFire had major stuttering/chopiness even if FPS were good on paper). It would be interesing comparisons if you could benchmark ports of the same game in different platforms with the defaults and the best optimizations available to see if my theory holds true, but it seems like a royal pain in the butt to prepare all that because you need in depth knowledge of a lot of stuff.

 

[igor_kavinski]

It would be interesing comparisons if you could benchmark ports of the same game in different platforms with the defaults and the best optimizations available to see if my theory holds true, but it seems like a royal pain in the butt to prepare all that because you need in depth knowledge of a lot of stuff.

Very good idea. That would be a very enlightening investigation, checking the performance characteristics of the same game with different executables, each for Windows, Linux and MacOS.

 

[igor_kavinski]

Intel's 14th Gen CPUs are discounted as low as $190, with up to $178 of free SSDs, coolers, RAM, or games tossed in — Is Intel cleaning out its closets?

More likely, Newegg is trying to move some older hardware with its CPUs, but it's hard to complain about a free SSD, RAM, cooler, or games.

www.tomshardware.com

I hope these deep discounts mean that Intel is getting ready to give a whole new meaning to buyer's remorse with the 12P gaming focused monster!

 

[nenforcer]

Intel's 14th Gen CPUs are discounted as low as $190, with up to $178 of free SSDs, coolers, RAM, or games tossed in — Is Intel cleaning out its closets?

More likely, Newegg is trying to move some older hardware with its CPUs, but it's hard to complain about a free SSD, RAM, cooler, or games.

www.tomshardware.com

I hope these deep discounts mean that Intel is getting ready to give a whole new meaning to buyer's remorse with the 12P gaming focused monster!

Yeah I was hoping we would here something official by now since Q3 has started but with the big summer sales currently ongoing I suspect we will have to wait a little longer. I am holding off on upgrading until I hear about the Barlett Lake chips. Any of the Core 5 8 core, the Core 7 10 core or 12 Core Core 9 will be an upgrade for me. The Core 5 120F slide circulating is definitely proof some new Core Series chips are incoming. I think the back to school season August is typically the big new sales push for the year they might get announced later this month for release in August / September.

 

[dullard]

I think the back to school season August is typically the big new sales push for the year they might get announced later this month for release in August / September.

I can't think of a single consumer-focused chip that was launched by Intel in July. There have been several June/August launches, but not July. And there have been no summer ones since Comet Lake. Recently, they have almost all been October through December launches (spilling over into CES early Jan sometimes).

Now, since Bartlett Lake is not really a consumer focused product, maybe it'll happen. I am more thinking these sales are in preparation for Arrow Lake Refresh.

 

[jpiniero]

Might be good to separate the launch of it with Arrow Lake Refresh.