Intel sneaks out nine 14th-Gen CPUs without E-cores — odd flagship Core i9-14901KE is the first overclockable chip for the embedded market
These new Raptor Lake Refresh chips are for embedded solutions only.
On the cusp of Intel's Arrow Lake-S desktop CPU launch, Intel has quietly launched an all-new 14th Gen Raptor Lake Refresh CPUs known as the "14001" series with P-cores only. These CPUs lack E-cores entirely and target embedded solutions, but in a strange twist, the flagship model is overclockable.
That's the bad news: these chips only come in an embedded format, preventing enthusiasts and mainstream customers from accessing these unorthodox CPUs. Embedded form factors are a computing solution where the CPU is installed into a machine from an OEM at the factory. Usually, these chips are sold with the motherboard and cannot be easily replaced or upgraded.
Intel's new E-coreless Raptor Lake Refresh lineup comprises three Core i9 SKUs, two Core i7 variants, and four Core i5 models. By far, the most unique variant of the nine models is the flagship Core i9-14901KE, which, if you couldn't tell by the model number, appears to be an overclockable K-series part. If this is true, the 14901KE is one of the only, if not the only, overclockable embedded processor that Intel has created to date.
Intel's new lineup is also very unorthodox in that it is the only lineup to date since 12th Gen that does not have any E-cores whatsoever. However, there are some positives with such a lineup. From Intel's side, having E-coreless chips allows it to recycle defective Raptor Lake dies that might have faulty E-cores but are perfectly working beside that. We suspect this is the sole reason why Intel decided to create this new lineup in the first place: to unload defective dies that probably have been gathering up since the 14th Gen launch.
Intel's new lineup is also advantageous for consumers. One of the main disadvantages of Intel's modern hybrid architectures is the additional complexity of scheduling work on suitable core clusters for optimal performance and power efficiency. Sometimes, OS schedulers will put some workloads on the wrong cores, causing performance headaches for the end user. Additionally, many workloads don't benefit from having E-cores; instead, they only benefit from having six or eight physical cores (or less).
All three Core i9 variants have eight Raptor Cove P-cores, 16 threads, and 36MB of L3 cache. The main differences between the three are clock speed and PBP/TDP. The flagship i9-14901KE has a 125W TDP, a maximum turbo boost clock of 5.8GHz, and a base clock of 3.8 GHz. The Core i9-14901E has a significantly lower 65W TDP but only loses 200 MHz off its maximum boost clock, featuring a peak turbo boost of 5.6 GHz. The base clock takes a significant hit, however, at 2.8 GHz. The i9-14901TE is the most power-constrained variant, featuring a 45W PBP/TDP. The peak turbo clock speed is 5.5 GHz, and the base clock is 2.3 GHz.
CPUs: | Cores / Threads | Clocks | Cache | TDP | Integrated Graphics Model |
---|---|---|---|---|---|
Core i9-14901KE | 8 / 16 | 5.8GHz Boost / 3.8GHz Base | 16MB L2 / 36MB L3 | 125W | UHD Graphics 770 / 32 EUs |
Core i9-14901E | 8 / 16 | 5.6GHz Boost / 2.8GHz Base | 16MB L2 / 36MB L3 | 65W | UHD Graphics 770 / 32 EUs |
Core i9-14901TE | 8 / 16 | 5.5GHz Boost / 2.3GHz Base | 16MB L2 / 36MB L3 | 45W | UHD Graphics 770 / 32 EUs |
Core i7-14701E | 8 / 16 | 5.4GHz Boost / 2.6GHz Base | 16MB L2 / 33MB L3 | 65W | UHD Graphics 770 / 32 EUs |
Core i7-14701TE | 8 / 16 | 5.2GHz Boost / 2.1GHz Base | 16MB L2 / 33MB L3 | 45W | UHD Graphics 770 / 32 EUs |
Core i5-14501E | 6 / 12 | 5.2GHz Boost / 3.3GHz Base | 12MB L2 / 24MB L3 | 65W | UHD Graphics 770 / 32 EUs |
Core i5-14501TE | 6 /12 | 5.1GHz Boost / 2.2GHz Base | 12MB L2 / 24MB L3 | 45W | UHD Graphics 770 / 32 EUs |
Core i5-14401E | 6 /12 | 4.7GHz Boost / 2.5GHz Base | 12MB L2 / 24MB L3 | 65W | UHD Graphics 730 / 24 EUs |
Core i5-14401TE | 6 / 12 | 4.5GHz Boost / 2GHz Base | 12MB L2 / 24MB L3 | 45W | UHD Graphics 730 / 24 EUs |
Very little differentiates the Core i9 models from the Core i7 variants. The two Core i7 SKUs also have eight cores and 16 threads, but they have slightly lower clock speeds and an extremely slight reduction in L3 cache. The Core i7-14701E has a 65W TDP, 5.4 GHz turbo frequency, 2.6 GHz base frequency, and 33MB of L3 cache. The Core i7-14701TE has the same core and cache configuration but features a 45W PBP/TDP, a max turbo clock of 5.2 GHz, and a base clock of 2.1 GHz.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
The Core i5 takes a much bigger hit to the spec sheet, featuring six cores, 12 threads, and 24MB of L3 cache. The Core i5 lineup is divided into two sub-lineups, the 14501 and 14401 series — much like the Core i5 14500 and i5 14400 desktop chips. Two main areas differentiate the Core i5 sub-lineups: the Core i5 14501 series comes with boost frequencies in the 5GHz range, while the 14501 series comes with peak turbo frequencies in the 4GHz range. The 14501 series is also the only group of chips in Intel's E-coreless 14th Gen lineup that uses the lower-end UHD 730 Graphics with 24 EUs. The rest of the i5s, i7s, and i9s use UHD 770 Graphics with 32 EUs.
Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.
-
watzupken I am thinking which brave person will buy it considering the existing issue with Raptor Lake processors that’s still ongoing now? Instead of providing their client an explanation and resolution, Intel double down on releasing new Raptor Lake SKUs.Reply -
bit_user I just want to add that Intel already did a similar thing with the Xeon E-2400 series.Reply
https://www.servethehome.com/intel-xeon-e-2400-series-brings-raptor-lake-to-servers/
Raptor Lake B0 dies, with E-cores disabled. -
greymaterial How bold of them to still label their chips "overclockable" when they can't even keep up their in-spec peak performance with reasonable stability. Are their marketing team living in a parallel universe?Reply -
usertests
It's possible that destroying the E-cores fixes the problem(s). Which would still be bad, but hey.greymaterial said:How bold of them to still label their chips "overclockable" when they can't even keep up their in-spec peak performance with reasonable stability. Are their marketing team living in a parallel universe? -
bit_user
No, disabling E-cores has already been tried. Doesn't fix it. In fact, it seems to help even less than limiting DRAM speeds.usertests said:It's possible that destroying the E-cores fixes the problem(s). Which would still be bad, but hey. -
usertests
It could be different than disabling them in software, if the problem is with specific portions of L3 cache on the die, or some other portion that would get completely disabled in these.bit_user said:No, disabling E-cores has already been tried. Doesn't fix it. In fact, it seems to help even less than limiting DRAM speeds.
All we really know for sure is that Intel has a big mess on their hands. -
bit_user
Given that disabling E-cores in BIOS will boost your ring bus frequency (on Alder Lake, at least), I doubt there's any further benefit to be had by fusing them off. That would seem to imply that disabling them in BIOS will also disable their L3 slices.usertests said:It could be different than disabling them in software, if the problem is with specific portions of L3 cache on the die, or some other portion that would get completely disabled in these.
Weirdly, Alder Lake-N has the L3 cache decoupled from the E-core clusters. I think you get 6 MB in all SKUs, regardless of how many cores are enabled. Look closely, right below the block labelled "Processor Transaction Router":
-
thestryker What's interesting about these parts when compared to the Xeons is that they still have all of the L3 despite not having the E-cores. Wonder if this is just binning leftovers from RPL/RPL-R because they seem like fairly odd parts.Reply -
m3city
Get out of your pro bubble man;). People are not aware of that particular issue, and don't give a crap about it. Even ones that build systems on their own. It has been scientifically proven, that most decisions we make, are basen on what our guts tell us, and if someone wants to build a new computer, is ready to spend $$, then one will assume that this new processor will be definitelly good, will run just fine (even if someone somewhere reported some issues - these happen, not here with me). Just look at how popular google and android are, despite beeing crap from user rights point of view. People don't care about that as long as they are provided with services that are perceived good and usefull.watzupken said:I am thinking which brave person will buy it considering the existing issue with Raptor Lake processors that’s still ongoing now? Instead of providing their client an explanation and resolution, Intel double down on releasing new Raptor Lake SKUs.
I personally (and guess you as well) will discourage any potential spender to put their money on intel right now. But our sphere of influence is limited - unless you drive purchase department in a company or advisory group. I tried to persuade colleguas at my company, guys who deal with Blender, but with no effect - they requested 14900K.
Edit: I find those embedded CPUs interesting. But I like ready to run, embedded solutions. -
zsydeepsky bit_user said:No, disabling E-cores has already been tried. Doesn't fix it. In fact, it seems to help even less than limiting DRAM speeds.
I saw some rumor suggests the CPU was unstable because they were oxidated. in another words, the cpu's metal circuits got rusty.
https://hardwaretimes.com/intel-13th-14th-gen-issues-manufacturing-defect-causing-oxidation-at-fault/
if that's the case, then it's guaranteed they will deteriorate further with time passes by.