Friday, June 27th 2025
Some Intel Nova Lake CPUs Rumored to Challenge AMD's 3D V-Cache in Desktop Gaming
Looking to challenge AMD's gaming CPU supremacy, Intel is reportedly developing Nova Lake processors with enhanced cache technology that could rival the popular 3D V-Cache found in X3D chips. According to leaker @Haze2K1, Intel plans to add "bLLC" (big Last Line Cache) to at least two Nova Lake models. This improved L3 cache is similar to AMD's 3D V-Cache, which has made X3D chips the top pick for enthusiast gamers since 2022. The new processors with bLLC will have 8 P-cores and 4 LP-E-Cores. One version will include 20 E-cores, while another will have 12 E-cores. Both are expected to keep a 125 W TDP rating.
Intel's bLLC technology already exists in Clearwater Forest server processors where local cache integrates into the base tile positioned beneath active tiles. This structural approach mirrors AMD's current 9000-series X3D design, where V-Cache attaches to the bottom of CPU dies—a significant improvement over earlier generations that placed cache on top, causing thermal issues and clock speed limitations. Yet, Intel said no to consumer plans for a technology similar to AMD's 3D V-Cache. In November 2024, Intel's Tech Communications Manager Florian Maislinger told YouTubers der8auer and Bens Hardware that they didn't plan such a desktop version. The Nova Lake-S family is set to hit the market in late 2026 or early 2027, with at least six desktop models using new LGA 1954 packaging. The lineup will start from the top-end Core Ultra 9 485K with 52 cores and 150 W TDP and go down to the basic Core Ultra 3 415K offering 12 cores at 125 W TDP.
Sources:
TechSpot, @Haze2K1 on X
Intel's bLLC technology already exists in Clearwater Forest server processors where local cache integrates into the base tile positioned beneath active tiles. This structural approach mirrors AMD's current 9000-series X3D design, where V-Cache attaches to the bottom of CPU dies—a significant improvement over earlier generations that placed cache on top, causing thermal issues and clock speed limitations. Yet, Intel said no to consumer plans for a technology similar to AMD's 3D V-Cache. In November 2024, Intel's Tech Communications Manager Florian Maislinger told YouTubers der8auer and Bens Hardware that they didn't plan such a desktop version. The Nova Lake-S family is set to hit the market in late 2026 or early 2027, with at least six desktop models using new LGA 1954 packaging. The lineup will start from the top-end Core Ultra 9 485K with 52 cores and 150 W TDP and go down to the basic Core Ultra 3 415K offering 12 cores at 125 W TDP.
140 Comments on Some Intel Nova Lake CPUs Rumored to Challenge AMD's 3D V-Cache in Desktop Gaming
I am a firm believer that even while slower in games than the X3D, the fact they got rid of Hyperthreading in the Ultra Series it aided a lot in overall stability, I did not experience a single stutter until I moved from my 285K to the 9950X3D. It felt more solid and snappier in Windows and I did not have to worry about setting CPPC in BIOS, making sure my cores are parking correctly, making sure games use the right cores, its just not something you can set and forget because once you do, a stutter here or there or a loss in fps will remind you. I will also never trust AMD to ever have solid drivers based on my collection of Radeon cards from 4850 to the 7900XTX and now stutterfest X3D CPUs (there's a dictionary online on what you can try to fix this but they usually miss). I would trade 200FPS off my total 700FPS in Valorant (for example) just to never experience a stutter again.
Is core parking not working? Is the stutter from crossing over CCD's? Is Intel's thread director just lightyears better preventing games from hitting ecores?
Core parking works, but you need to make sure it works and is working by enabling game mode, setting power profiles to balanced etc. Non-enthusiasts wouldn't know that they've been playing games on the wrong CCD lmao. Regarding stuttering, nobody knows where its coming from there's just a bunch of band-aid fixes online but I am betting it's due to a combination of AMD's amazing drivers, X3D Cache and Hyperthreading.
Its a waste of money to put it on low end chips, to begin with.
I've never had any stuttering issues on my AMD system though, sounds like a stability issue.
hardwaretimes.com/intel-15th-gen-cpus-to-get-rentable-units-why-hyper-threading-is-going-away/
Besides, entire line-up does not need extra cache. Those SKUs are specifically for gaming crowds.
Intel just had to increase the L3 cache size, especially with Meteor Lake where they have the packaging capability to just create an L3 tile and add it to the mix. Hell, some Arrow Lake SKUs have blank "filler" tiles next to the compute tile to keep the package rectangular, yet they top at out 36MB L3 cache against AMD's 128MB. Intel have the manufacturing capability and flexibility, as well as spare physical space to add L3 cache to their products without touching AMD's patent. They just didn't.
If Intel can compete in gaming loads again then it's good for everyone. I'm just looking at the dumpster-fire that is the Arrow Lake gaming benchmarks where they're often slower than 13th gen with a few popular titles their average FPS is worse than AMD's minimum FPS. Not a good look for Intel and gaming.I like e-cores, just not for gaming, for the same reason I love laptops with Zen C cores.
Since Alder Lake, Intel have had 2P8E U-series processors that make for very capable productivity laptops where the workloads either need burst single-threaded performance, or they need maximum multi-threaded performance per Watt. A cluster of 4 E-cores running at ~2.5GHz does way more rendering per Watt, for example, than a single P-core running at 4GHz+ That same die area could have instead been a 4P0E processor and we all know how quad cores suck at multi-threaded workloads these days ;)
I think their aproach with previous arch was more general purpose.
Efficiency and reliability aside, the 14900k was very good in gaming as well.
High volume, mass production requires factories to remain at peak capacity or the profitability of the factory goes into the red. A drop in production by just a little can lead to a factory closing. This is where Intel finds itself. Instead of 10 million chips per year (made up number) they are only making 7.5 million chips (made ip number). Even though revenue is still many billions, profits are negative.
You can only last that long for a few quarters. Intel has been losing money per quarter for the better part of a year now. All the lucrative government, corporate and server contracts will not be enough to get Intel fabs back to peak capacity again in such a crowded chip market.
So you tell me which is better, TSMC that is close to breaking $30B per quarter or Intel that is close to dropping below $10B per quarter? I’ll give you a hint (whispers in ear) the higher number is better.
To me the ones that count are mid and higher, low end means nothing to me.