Monday, February 23rd 2026

Intel Plans Return to Unified Core Design, No More Performance and Efficiency Core Split

According to the latest job listings, we learn that Intel is planning the return of a unified core architecture, something we haven't been accustomed to in the last few years. Starting with the 12th Generation "Alder Lake" processors, Intel began selling hybrid core processors that combine "Golden Cove" performance cores and "Gracemont" efficient cores. These became commonly known as P and E-Cores, which are now being used across Intel products in hybrid designs or P/E-core-only Xeon processors, of course, with the latest designs and iterations. However, as the new job listings suggest, Intel is assembling a team of engineers for its "Unified Core" design group that will deliver the new microarchitecture to power the next generation of processors.

Separating the core design into P-Cores and E-Cores has yielded Intel the desired results that the company hoped for. This includes product separation and many goals across platforms. For example, in the consumer sector, E-Cores run a lot of side and background tasks in the operating system, while P-Cores power main applications like games. To extract maximum performance, Intel has a dedicated Thread Director that makes the entire process work and dictates just what application goes to which core, in tandem with the operating system. Intel also provides P-Core-only and E-Core-only Xeon server processors that serve either performance sectors like HPC and AI or the cloud sector that needs many cores with somewhat lower performance, but in a dense 100-core+ package.
With a unification of the P-Core and E-Cores, Intel would have to look into other techniques for separating its product offerings. That can simply be done by using smaller cache capacities, as L2 and L3 cache occupy a huge percentage of the CPU die, just like AMD does with Zen 5 and Zen 5c. We can hope to see other ways, as Intel's Unified Core design team could bring surprises and some interesting new technologies. The timeline for the launch of this "Unified Core" is still unknown, as the design is probably in the early stages. These processes take years, and it is likely that this new core design will be launching by the end of the decade. Until then, we are looking into the hybrid core design like we have been using in recent years.
Source: LinkedIn Job Posting
Add your own comment

127 Comments on Intel Plans Return to Unified Core Design, No More Performance and Efficiency Core Split

#1
RejZoR
Different core clusters work on paper, but not in practice. I don't think this will ever take off on desktops. Smartphones yes, because it's pretty well defined what's foreground, what's background and what performance each needs. It's not that simple in Windows.
Posted on Reply
#2
Nostras
Could be that Intel sees P + LPE as the future. Or perhaps they want to go the AMD route and introduce something like Pc cores. Heh.
Posted on Reply
#3
lambda
I think it's a reach to conclude or even suspect that "Unified Core" here refers to a singular core type based design.

For all we know it could be a business moniker.

Nothing in that screenshot points to any specific details, the word "unified" is performing the heavy lifting here.

Edit: Not directed at the author of the article.
Posted on Reply
#4
N/A
It's probably a derivative of Ecore with 5-6 ALu. Now an E-core with 4 ALu takes up a similar amount of space on the chip as 2.5 MB L2, but reducing it to 128K for segmentation had a terrible effect on Celerons in the past. I'd rather have 16 E-cores sharing all 16MB L2, rather than each 4 LPE island having its own 4MB.
Posted on Reply
#5
qcmadness
N/AIt's probably a derivative of Ecore with 5-6 ALu. Now an E-core with 4 ALu takes up a similar amount of space on the chip as 2.5 MB L2, but reducing it to 128K for segmentation had a terrible effect on Celerons in the past. I'd rather have 16 E-cores sharing all 16MB L2, rather than each 4 LPE island having its own 4MB.
The major issue for e-core is not caching but rather OOO executions.
Posted on Reply
#6
Mack4285
Didn't they start the P/E cores to increase core count to look competitive with AMD Zen 8 cores, but retain a non-crazy TDP (since they were stuck at 14+++ nm)? And now that they think they have the TDP under control again, they basically want to go back to only P cores. That's what it looked like to me. AMD has said they don't need this separation of cores, and same goes for Qualcomm Snapdragon Elite.
Posted on Reply
#7
kondamin
lambdaI think it's a reach to conclude or even suspect that "Unified Core" here refers to a singular core type based design.

For all we know it could be a business moniker.

Nothing in that screenshot points to any specific details, the word "unified" is performing the heavy lifting here.

Edit: Not directed at the author of the article.
That's pretty much what I'm thinking.
wouldn't be that surprising to see 'p' cores go away as e cores get more functionality
Posted on Reply
#8
cchi
Could be that Intel sees P + LPE as the future. Or perhaps they want to go the AMD route and introduce something like Pc cores. Heh.
It will likely be BE-cores + E-cores + LPE cores.
Instead of Pc it will be BE (bigger E cores), as current P-cores are not much better than E-cores already.
A boosted E-core design tuned for higher frequency and bigger caches should be able to replace the P-core performance point.
So one base core design, with several variants.

The E-core design is newer so it likely has less technical debt, and therefore can easier scale IPC going forward.
Apple has proven that IPC can be still improved a lot, so it should make sense that they want to catch up as much as they can on that front using a similar approach.
Posted on Reply
#9
Veseleil
Mack4285AMD has said they don't need this separation of cores, and same goes for Qualcomm Snapdragon Elite.


They are still separated afaik...
Posted on Reply
#10
Simon.J
lambdaI think it's a reach to conclude or even suspect that "Unified Core" here refers to a singular core type based design.

For all we know it could be a business moniker.

Nothing in that screenshot points to any specific details, the word "unified" is performing the heavy lifting here.

Edit: Not directed at the author of the article.
I suspect that Intel is looking to make similar core as in IBM Z systems where a core has more than two threads, I am not an engineer but it looks like it would be similar or same approach.
Posted on Reply
#11
Dammeron
RejZoRDifferent core clusters work on paper, but not in practice. I don't think this will ever take off on desktops. Smartphones yes, because it's pretty well defined what's foreground, what's background and what performance each needs. It's not that simple in Windows.
It works on smartphones cause they need as much power efficiency as possible. Desktops can handle a bit more power draw and heat without a problem though.
Posted on Reply
#12
Simon.J
Veseleil

They are still separated afaik...
Well, I assume a key feature here is Intel’s APX ISA, which many pointed out that would get architecture closer to RISC as in IBM Z where a massive core has more than 2 threads.
Posted on Reply
#13
Tek-Check
cchiIt will likely be BE-cores + E-cores + LPE cores.
Instead of Pc it will be BE (bigger E cores), as current P-cores are not much better than E-cores already.
A boosted E-core design tuned for higher frequency and bigger caches should be able to replace the P-core performance point.
So one base core design, with several variants.
This.
cchiThe E-core design is newer so it likely has less technical debt, and therefore can easier scale IPC going forward.
Apple has proven that IPC can be still improved a lot, so it should make sense that they want to catch up as much as they can on that front using a similar approach.
I agree. P cores need debloating due to legacy technical debt whereas E cores need that Ooomph! factor and HT back on the table.
Posted on Reply
#14
qcmadness
Simon.JI suspect that Intel is looking to make similar core as in IBM Z systems where a core has more than two threads, I am not an engineer but it looks like it would be similar or same approach.
It depends on how much back-end resources that could be used by extra threads. Security-related issues are there, too.

Historically Intel CPU are more starved in back end while AMD CPUs are more starved in front end.
Posted on Reply
#16
Dirt Chip
Yes, just like they abandoned HT for good after RL refresh (14th gen).
Posted on Reply
#17
Lycanwolfen
Drop the E cores and P core and just make a pure CORE. They have lost market share because AMD makes pure core cpus. We never wanted 8 P core 16 Ecore. We wanted a 16 core cpu period or 32 core or for ppl that can afford it those threadrippers 192 core.

Too bad we don't have a quantum based x86 unlimited core.
Posted on Reply
#18
Onasi
LycanwolfenDrop the E cores and P core and just make a pure CORE. They have lost market share because AMD makes pure core cpus. We never wanted 8 P core 16 Ecore. We wanted a 16 core cpu period or 32 core or for ppl that can afford it those threadrippers 192 core.
Me when I pretend like Zen C doesn’t exist. Intel can just go the same route and make both a “power” oriented core and an “efficiency” oriented one on the same ISA.
Posted on Reply
#19
ThomasK
Proof that all those e-cores were only a marketing and PR gimmick, to get the uninformed consumer into thinking "more is better".
Posted on Reply
#20
Youseethis
The question is, will they keep as base the p cores, the e cores or make a new design?
Posted on Reply
#21
Squared
Mack4285Didn't they start the P/E cores to increase core count to look competitive with AMD Zen 8 cores, but retain a non-crazy TDP (since they were stuck at 14+++ nm)? And now that they think they have the TDP under control again, they basically want to go back to only P cores. That's what it looked like to me. AMD has said they don't need this separation of cores, and same goes for Qualcomm Snapdragon Elite.
I thought E cores were brought to desktop to compete with chiplets. Comet Lake-S brought 10 cores in the 10900K, Rocket Lake being backported from 10SF to 14nm+++++ brought only 8 cores in the 11900K, and so Alder Lake on Intel 10ESF/Intel 7 would have 10 cores again. Trouble was that Comet Lake ended up competing not with just the 3800X but also the dual-CCD 3950X. Intel would have no answer to chiplets until Meteor Lake. So to help Alder Lake-S compete, the heterogenous concept tested in Lakefield was used, replacing 2 Golden Cove cores with 2 Gracemont core clusters. The resulting 16 cores compete very well against the Ryzen 5950X, even in power efficiency.

But the Gracemont cores aren't power-efficient on desktop; they are clocked too high and are actually less lower efficient than the Golden Cove cores. Like AMD's chiplets, they lower manufacturing costs. Meteor Lake addressed this with Crestmont cores which improved more than the Redwood Cove cores did.
Posted on Reply
#22
docnorth
Mack4285Didn't they start the P/E cores to increase core count to look competitive with AMD Zen 8 cores, but retain a non-crazy TDP (since they were stuck at 14+++ nm)? And now that they think they have the TDP under control again, they basically want to go back to only P cores. That's what it looked like to me. AMD has said they don't need this separation of cores, and same goes for Qualcomm Snapdragon Elite.
Mostly agree, although it started at 10nm. Anyway the hybrid design imho will continue in mobile chips, be it p+e cores or one type of cores with different clocks. But even with the improved efficiency the latter approach might become more prominent on desktop too, 2 p(?) cores could clock much higher than the others, not only 100-200 MHz.
Veseleil

They are still separated afaik...
This. And Intel is already (marginally) using this approach, 2 p cores clock -for now- only 100-200 MHz higher.

In any case improved efficiency remains the key, it provides many options, otherwise…:slap:
Posted on Reply
#23
igormp
cchiIt will likely be BE-cores + E-cores + LPE cores.
Instead of Pc it will be BE (bigger E cores), as current P-cores are not much better than E-cores already.
A boosted E-core design tuned for higher frequency and bigger caches should be able to replace the P-core performance point.
So one base core design, with several variants.

The E-core design is newer so it likely has less technical debt, and therefore can easier scale IPC going forward.
Apple has proven that IPC can be still improved a lot, so it should make sense that they want to catch up as much as they can on that front using a similar approach.
Wasn't that the plan since the beginning? I remember seeing talks about the E-cores evolving until they could entirely fufill the P-core duties, making it easy to take the AMD route of having higher-clocking variants with more cache, and more dense variants of the same core with lower clocks, less cache and maybe halved SIMD units for way better power efficiency.
Simon.JI suspect that Intel is looking to make similar core as in IBM Z systems where a core has more than two threads, I am not an engineer but it looks like it would be similar or same approach.
They did have the whole talks of rentable units and whatnot back then. But I don't think they would go with 4 or 8-way SMT in its current form, their front-end is capable enough of making the backend busy already.
qcmadnessIt depends on how much back-end resources that could be used by extra threads. Security-related issues are there, too.

Historically Intel CPU are more starved in back end while AMD CPUs are more starved in front end.
The current E-cores have a hella interesting front-end, with that 3x 3-wide decode cluster, more than what Apple had with the M1 (8-wide), and close to the M4 (10-wide).
I wonder if adding a µop cache to it would improve it further, since Zen 5 has shown that (as long as the dual 4-wide decoders are enabled) its impact is not huge. Nonetheless, skymont/darkmont is clearly starved with their backend, specially when it comes to branch prediction. Feeding it a larger ROB (and load/store queues as well), and managing to increase the dispatcher may bring it really close to AS-levels, with a really wide core that doesn't need to clock as high.
Posted on Reply
#24
Harun8
I do think an approach more like Zen 5 and Zen 5c is better than P / E. Less engineering resources needed. But I do like having the different core types. They are useful for servers/VMs where high core counts are extremely beneficial, even if they're weaker cores
Posted on Reply
#25
LastDudeALive
RejZoRDifferent core clusters work on paper, but not in practice. I don't think this will ever take off on desktops. Smartphones yes, because it's pretty well defined what's foreground, what's background and what performance each needs. It's not that simple in Windows.
Apple has P and E cores in their laptop and desktops. And even if these big/little designs that AMD and Intel have didn't exist, Windows and Linux would still assign priority to different tasks, because there's lots of other things that can affect performance besides core architecture. High-priority threads get assigned to the highest performing cores (even if all the cores are the same architecture), they get their data cached in RAM, etc. And AMD's chiplet design presents similar challenges. In dual-CCD models, Windows needs to keep high-priority threads on 1 CCD and background tasks on the other to maximize performance.

And it's not just to divide high/low priority tasks. Efficiency is a huge reason for AMD/Intel's big/little designs. And that's very important in the laptop market, where they are currently getting manhandled by Apple. It's also important in the enterprise market (which is more important than the consumer market altogether). Enterprise customers loooooove the dense e-core only Xeons and AMD's c-core only EPYCs. And it's important to keep development costs under control by having similar architectures for enterprise and consumer.
Mack4285Didn't they start the P/E cores to increase core count to look competitive with AMD Zen 8 cores, but retain a non-crazy TDP (since they were stuck at 14+++ nm)? And now that they think they have the TDP under control again, they basically want to go back to only P cores. That's what it looked like to me. AMD has said they don't need this separation of cores, and same goes for Qualcomm Snapdragon Elite.
AMD has separation of cores, with their c cores. They have taken a different approach to it, opting to keep the same architecture instead of designing a new one, but it has the exact same limitations as Intel's E cores. Since they're significantly slower and have half the L3 cache, you need the OS to keep high-priority tasks (like gaming) off the c cores and on the P cores.

It was originally a response to AMD's node/core count advantage, yes. It wasn't just to "look" better, it did match performance. And now that they're on similar nodes, we can see that the density of single-threaded e-cores enables the Core Ultra CPU to match multi-threading power with lower thread counts. i.e. small cores are better for multi-threading, which, of course, is also shown by the success of e-core/c-core Xeon/EPYC designs.
DammeronIt works on smartphones cause they need as much power efficiency as possible. Desktops can handle a bit more power draw and heat without a problem though.
Laptops also need power efficiency. They're a very important market, and AMD and Intel need to compete against the king of efficiency, Apple. And sure, desktops can handle more, but Intel also deservedly got roasted (pun intended) with their 400W+ 13900K/14900K models and the overall power consumption of all Raptor Lake models. And remember that efficiency = power. If a design is more efficient, that means that when it gets more power in a desktop, you get a lot more multithreading power out of it.
ThomasKProof that all those e-cores were only a marketing and PR gimmick, to get the uninformed consumer into thinking "more is better".
If a "marketing gimmick" can essentially match productivity performance with 8 less threads, I'll take more marketing gimmicks, please.
Posted on Reply
Add your own comment
Nov 18th, 2026 04:25 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts