Friday, January 30th 2026

AMD "Zen 6" CCD Packs 12 Cores, 48 MB L3 Cache

The standard CPU complex die (CCD) chiplet based on AMD's next-generation "Zen 6" microarchitecture is poised to feature a 50% increase in CPU core counts. This would be the first time that AMD has increased core-counts of its CCDs containing full-sized CPU cores capable of sustaining high clock speeds. It had developed CCDs with high core counts using compacted cores with lower clock speed limits until now. The standard "Zen 6" CCD is expected to come with a similar die-size to the current "Zen 5" CCD, with a rumored die size of 76 mm², which is slightly larger than the 71 mm² of the "Zen 5" CCD.

The "Zen 6" CCD is expected to pack 12 CPU cores, all part of a single CPU core complex (CCX). All twelve cores share a 48 MB L3 cache. Both the CPU core counts and L3 cache size hence see a 50% increase over "Zen 5." AMD is building the "Zen 6" CCD on TSMC N2 (2 nm nanosheet) process node, which evidently offers a massive increase in transistor densities over the current TSMC N4P (4 nm FinFET) on which the "Zen 5" CCD is made. With an eye on Intel's pivot towards a cache rebalancing in favor of large last-level caches on some of its premium Core Ultra 400 "Nova Lake-S" desktop processors, AMD is expected to give the "Zen 6" CCD full 3D V-Cache capability, and the company's next X3D processors could come with L3 caches of 144 MB per CCD for up to 288 MB of L3 cache on the desktop AM5 socket.
Sources: HXL (Twitter), VideoCardz
Add your own comment

99 Comments on AMD "Zen 6" CCD Packs 12 Cores, 48 MB L3 Cache

#76
melkor_unlimited
AssimilatorZen6 will have at best an extra 4 PCIe 4.0/5.0 lanes, and those will be exclusively reserved for the ASM4242 USB4 chipset, because AMD has figured out that without any competition they can screw consumers just as much as Intel used to.
On which pins would you connect those PCIe lines?
upload.wikimedia.org/wikipedia/commons/2/2d/Socket_AM5_pinmap.svg
NociThe average consumer probably not, but here with all the enthousiast there can never be enough :D
With some flexibility in using the lanes the amount of them would not be so restrictive.
Posted on Reply
#77
AusWolf
NociThe average consumer probably not, but here with all the enthousiast there can never be enough :D
So how much does "an enthusiast" need? Who is an enthusiast anyway? :D
Posted on Reply
#78
CornyOC
ChaneI wonder if Zen 6 will have the new Level 2 cache design AMD has been working on too or maybe that's a Zen 7 thing. Anyway, I'm excited to get more cores in a single CCD without doing bigLITTLE.
all that was said that they are working on stacked L2 cache design, I hope it will be implemented with Zen 6
Posted on Reply
#79
Assimilator
AusWolfDo consumers even need more PCI-e lanes?
Considering that USB4 takes 4, yes. That's an entire SSD of bandwidth, versus Intel where USB4 is built-in and thus costs no lanes.

But really, what would alleviate Zen 4/5/6's lane shortage would be if AMD would just allow board manufacturers to allocate and bifurcate those lanes. So for example, Instead of forcing x16 GPU (slot) + x4 SSD +x4 SSD, a manufacturer should be able to offer a board with triple slots that each run at x8.
Posted on Reply
#80
AusWolf
AssimilatorConsidering that USB4 takes 4, yes. That's an entire SSD of bandwidth, versus Intel where USB4 is built-in and thus costs no lanes.
I'm not sure if a consumer needs USB 4 as standard anno 2026. Give it another 2-4 years, and maybe I'll agree.
AssimilatorBut really, what would alleviate Zen 4/5/6's lane shortage would be if AMD would just allow board manufacturers to allocate and bifurcate those lanes. So for example, Instead of forcing x16 GPU (slot) + x4 SSD +x4 SSD, a manufacturer should be able to offer a board with triple slots that each run at x8.
Or to sacrifice the second m.2, or the PCI-e x4 slot for lanes for another device, like a USB 4 controller chip instead.
Posted on Reply
#81
Punkenjoy
Having a chipset connected via PCI-E 4X that provide additional PCI-E lanes, USB4, etc is totally fine for 95% of the cases. Indeed in some copy scenario, it might affect copy speed but for most consumer grade desktop devices, its totally fine. If you needs all those lanes, Threadripper platform is there.

Intel and AMD build their consumer platform with consumer in mind, not professional. They have professional platform that will provide what you needs. Some might think they need professional platforms for their daily usage, but they just have to buy the professional platform.

It always have been like that
Posted on Reply
#82
evernessince
AssimilatorBut really, what would alleviate Zen 4/5/6's lane shortage would be if AMD would just allow board manufacturers to allocate and bifurcate those lanes. So for example, Instead of forcing x16 GPU (slot) + x4 SSD +x4 SSD, a manufacturer should be able to offer a board with triple slots that each run at x8.
AMD does allow that. There are AM5 boards with 3 slots (although only 1 that actually runs 3 at x8 in the MSI Godlike. Cheaper options with the 2nd and 3rd slots running at PCIe 4.0 and PCIe 3.0 exist as well). End users can bifurcate the slots as well if you wanted to install more devices than slots available. AGESA supports this natively on AM5, it's just up to the board partner to implement it. I know ASUS gates it off to high-end board but ASRock allows it across all their boards.

The reason you don't see three x8 much is cost / benefit. The vast majority of people just stick a GPU in there and are done slot wise. Both PCIe 4.0 and 5.0 require a repeater and 5.0 requires more layers as well. The requirements will get even worse with PCIe 6.0.
Posted on Reply
#83
PenguinBelly
MufasaVNvidia more than 10X their L2 cache in one Gen. That was after AMD did the same with L3 cache. GPUs. CPUs are the sane issue for AMD X3D. Intel are the only ones who haven't tied it in any product and this stand the most to gain from it despite trade offs
Intel did it with Broadwell.
Posted on Reply
#84
efikkan
AssimilatorThat's why I said at best. AM5 was an anemic socket at launch and Zen6 won't change that, actual power users will almost certainly have to wait until the next socket for any real IO updates. That of course of course assumes AMD will even develop a new socket, or just use AM5 forever like they did with AM4 and subject us to even more years of mediocrity.
Perhaps if the four lanes to the chipset is upgraded to PCIe 5.0 speeds there would be some increased usability. But I've seen nothing to indicate this so far.

As for the discussion of what people need, pure gamers can do fine on AM5, but they better choose one of the few motherboards which allows those 4 lanes reserved for USB 4 to be used for an additional M.2 or PCIe slot, as otherwise they will be left with only one SSD at full speed and one (or more) sharing bandwidth through the chipset. Pretty much any power user would at least want 3 "full speed" SSDs; OS, games and work drive, and on AM5 one of those will in practice have to run at "half speed". With higher tier boards coming for the next gen we can expect motherboards to feature 10G NICs, which alone will suck up 1 lane from the chipset, which further emphasizes the need for bandwidth.

And while splitting the x16 slot is fine for some, we are talking about prosumers here, and assuming someone would want to upgrade the GPU one or two generations from now when GPUs will be PCIE 6.0, having those x16 lanes for the GPU alone will become important.
Posted on Reply
#85
_roman_
when wendel, which i usally do not consume, praise supermicro and pcie 6 on expensive servers - I personally highly doubt to see pcie 6 soon on consumer motherboards.

the am5 chipset / hub sharing is a cost feature and on purpose. i do not expect that to be solved.
it is nice to not see pcie 3 ssds anymore on am5. the main reason for my am5 transition. 1x pcie 3x4 amd 1x pcie 4.0 was too slow. multiple pcie 4.0x4 are fine.

why else would someone not buy threadripper?
Posted on Reply
#86
Assimilator
_roman_why else would someone not buy threadripper?
Because it's ridiculously overpriced for what you get. 2 bajillion cores and motherboards pimped out to the max with features I'll never use is not "prosumer" but rubbish. I'd happily buy EPYC CPUs, except the boards for those are so barebones as to make them pointless, especially since ASM4242 add-in cards still require proprietary GPIO headers - meaning getting USB4 via an AIC is effectively impossible.
Posted on Reply
#87
melkor_unlimited
AusWolfSo how much does "an enthusiast" need? Who is an enthusiast anyway? :D
Enthusiast - somebody who wants to work at professional level but does not want to pay for professional tools.
AssimilatorConsidering that USB4 takes 4, yes. That's an entire SSD of bandwidth, versus Intel where USB4 is built-in and thus costs no lanes.

But really, what would alleviate Zen 4/5/6's lane shortage would be if AMD would just allow board manufacturers to allocate and bifurcate those lanes. So for example, Instead of forcing x16 GPU (slot) + x4 SSD +x4 SSD, a manufacturer should be able to offer a board with triple slots that each run at x8.
Three x8 slots?
Throw off that crappy Prom21 and you can get that. Three x8 at Gen5 speed.
+ 1x 480 Mbps USB 2.0
+ 4x 10 Gbps USB 3.2 Gen 2
- no SATA
- no NVME

Edit: 28 lanes total = 3 groups of x8 + one x4 to chiplet
Sorry. Was confused by TPU database showing 24 lanes. Its 24 usable lanes after connecting chiplet.
PunkenjoyHaving a chipset connected via PCI-E 4X that provide additional PCI-E lanes, USB4, etc is totally fine for 95% of the cases. Indeed in some copy scenario, it might affect copy speed but for most consumer grade desktop devices, its totally fine. If you needs all those lanes, Threadripper platform is there.
There is a "little" issue. Prom21 uses 4 lanes. But work only at Gen4 speed (half of bandwidth that Ryzen 7000/9000 provides).
efikkanPerhaps if the four lanes to the chipset is upgraded to PCIe 5.0 speeds there would be some increased usability. But I've seen nothing to indicate this so far.
On the motherboards already providing Gen5 PCIe lanes to GPU, NVME, USB4 it would not increase cost (so much) for another 4 lanes to chipset. Even if chipset itself would provide only Gen4 lanes.
But you have to ask AsMedia when they will condescend to learn how to use Gen5 in their chips.
efikkanAs for the discussion of what people need, pure gamers can do fine on AM5, but they better choose one of the few motherboards which allows those 4 lanes reserved for USB 4 to be used for an additional M.2 or PCIe slot, as otherwise they will be left with only one SSD at full speed and one (or more) sharing bandwidth through the chipset.
Older ones (600 series) where USB4 was not mandatory.
Or some of the few 800 series which allow user to chose where they want use those Gen5 lanes.
efikkanPretty much any power user would at least want 3 "full speed" SSDs; OS, games and work drive, and on AM5 one of those will in practice have to run at "half speed".
Without sacrificing lanes from GPU:
AM5 - 2 NVME slots Gen5 x4
s1700 - zero
s1851 - 2 NVME slots Gen5 x4

Any more and you use lanes from GPU or "half-speed" lanes from chipset.
efikkanAnd while splitting the x16 slot is fine for some, we are talking about prosumers here, and assuming someone would want to upgrade the GPU one or two generations from now when GPUs will be PCIE 6.0, having those x16 lanes for the GPU alone will become important.
With enough of VRAM at GPU the difference between PCIe Gen speed is marginal.
All of those miners used GPU connected via x1 Gen3.
Posted on Reply
#88
Assimilator
melkor_unlimitedThree x8 slots?
Throw off that crappy Prom21 and you can get that. Three x8 at Gen5 speed.
+ 1x 480 Mbps USB 2.0
+ 4x 10 Gbps USB 3.2 Gen 2
- no SATA
- no NVME
melkor_unlimitedThere is a "little" issue. Prom21 uses 4 lanes. But work only at Gen4 speed (half of bandwidth that Ryzen 7000/9000 provides).
Because AMD can design a CPU with a PCIe 5.0 IO die, but apparently making that same IOD a standalone chipset and bolting on a few bits like USB and SATA is so difficult that they had to farm it out to ASMedia. To be precise, it's not that it's difficult - it's that AMD chose to screw over consumers with an intentionally inferior chipset from launch day of Zen4.
melkor_unlimitedBut you have to ask AsMedia when they will condescend to learn how to use Gen5 in their chips.
I don't understand why companies are apparently incapable of designing their products to autonegotiate to the appropriate PCIe version and lane count to their peripheral. Your device needs 4 lanes of PCIe 4.0 bandwidth? Cool, if I plug it into a PCIe 5.0 slot it should work at full speed and consume only 2 lanes, not hog 4 because it's hardcoded to PCIe 4.0 lane speeds.
Posted on Reply
#89
LastDudeALive
AssimilatorI don't understand why companies are apparently incapable of designing their products to autonegotiate to the appropriate PCIe version and lane count to their peripheral. Your device needs 4 lanes of PCIe 4.0 bandwidth? Cool, if I plug it into a PCIe 5.0 slot it should work at full speed and consume only 2 lanes, not hog 4 because it's hardcoded to PCIe 4.0 lane speeds.
Because lanes are physical. If a PCIe slot has 4 lanes, that's 4 wires running directly from the slot to the CPU pins. Those wires cannot be rewired to go somewhere else if the add-in card only needs 2 of them.
Posted on Reply
#90
efikkan
melkor_unlimitedWith enough of VRAM at GPU the difference between PCIe Gen speed is marginal.
Running a higher tier GPU from one or two generations ahead at x16 5.0 speed will likely be fine, but with only x8 lanes and 5.0 speed, a 6.0 card will only have a quarter of the designed bandwidth, at that point it will likely matter.
AssimilatorBecause AMD can design a CPU with a PCIe 5.0 IO die, but apparently making that same IOD a standalone chipset and bolting on a few bits like USB and SATA is so difficult that they had to farm it out to ASMedia. To be precise, it's not that it's difficult - it's that AMD chose to screw over consumers with an intentionally inferior chipset from launch day of Zen4.
At this point I would prefer if we got rid of the south bridges and just got PCIe lanes instead, as it would cut down the cost a lot.
AssimilatorI don't understand why companies are apparently incapable of designing their products to autonegotiate to the appropriate PCIe version and lane count to their peripheral. Your device needs 4 lanes of PCIe 4.0 bandwidth? Cool, if I plug it into a PCIe 5.0 slot it should work at full speed and consume only 2 lanes, not hog 4 because it's hardcoded to PCIe 4.0 lane speeds.
Well, if it's developed in accordance with the PCIe 5.0 spec, then that will be doable, albeit more expensive to develop and validate. If it's developed before the spec is finalized, then it's a different story.
Posted on Reply
#91
Pets4Ever
Gotta love the passive-aggressive 🤣 likes when people assume certain things.

Yes, I referenced gaming performance on a high core count part. It's not that think having 12 or 24 cores is going to make a huge difference in most games, I just want high core counts for other applications without having to take a hit to gaming performance.
Posted on Reply
#92
AusWolf
melkor_unlimitedEnthusiast - somebody who wants to work at professional level but does not want to pay for professional tools.
Where did you get that from? What does that even mean?
Posted on Reply
#93
Drash
melkor_unlimitedEnthusiast - somebody who wants to work at professional level but does not want to pay for professional tools.
And the opposite also applies - professional wants enthusiast board because he's already blown £65,000 for this year's software licences and he hopes he can get away with just 2 channel 128GB RAM rather than 8 channel b/w, as the boss is thinking about downgrading the coffee brand to save a quid, FFS. We all want more for less, it's human. ;)
Posted on Reply
#94
randomdude9182
ARFGreed will lead you nowhere - especially when AMD is about to become the underdog once again after the release of the new 52-core Intel CPUs. AMD will have no answer, except lower pricings.



Is 6000-6400 too low for DDR5? It must be DDR5-8000, at least.

I won't upgrade with any AMD CPU in the distant future. I have my 24-thread Ryzen 9 5900X and it is very fast in its rated TDP of just 105W.
'Greed will lead you nowhere - especially when AMD is about to become the underdog once again after the release of the new 52-core Intel CPUs. AMD will have no answer, except lower pricings."
The 52 core will consume ~400W and be threadwripper priced (doubling core count and going from 3->2nm when your best chip consumes 253W under multithread, 379.5 and thats expecting them to NOT raise clocks and have perfect scaling).
Posted on Reply
#95
LastDudeALive
randomdude9182'Greed will lead you nowhere - especially when AMD is about to become the underdog once again after the release of the new 52-core Intel CPUs. AMD will have no answer, except lower pricings."
The 52 core will consume ~400W and be threadwripper priced (doubling core count and going from 3->2nm when your best chip consumes 253W under multithread, 379.5 and thats expecting them to NOT raise clocks and have perfect scaling).
Threadripper priced is probably an overstatement. A significant portion of Threadripper/Xeon's cost and value comes from the expanded platform. Quad/Octa channel memory, 80+ PCIe lanes, etc. Desktop won't have that. It will probably be ~$700+, but so will AMD's 24c/48t Zen 6 models, especially the X3D variants.
Posted on Reply
#96
randomdude9182
Intel said it was a "half way house" between hedt (2000$+) and the ultra high end. It's also using a shit to of sillicon, way more than amd's 24 core. It is going to cost at least 1200$ or else Intel is bleeding money.
Posted on Reply
#97
Fizzypop
I can see 2 chiplet offerings , the 12 core which is getting most of the headlines and a 8 core 3nm version for the R3 6 core and 8 core. The gaming champ will vary game to game depending on what they are optimized for, initially the 8 and 10 cores will be top dogs but as the years go by the top dog core count will go up. The other option will be to have an Apu offering for the six and eight core offerings , one chip 2 markets and less waisted silicon.
Posted on Reply
#98
AusWolf
FizzypopI can see 2 chiplet offerings , the 12 core which is getting most of the headlines and a 8 core 3nm version for the R3 6 core and 8 core. The gaming champ will vary game to game depending on what they are optimized for, initially the 8 and 10 cores will be top dogs but as the years go by the top dog core count will go up. The other option will be to have an Apu offering for the six and eight core offerings , one chip 2 markets and less waisted silicon.
I doubt that AMD will develop two different CCDs for one generation, especially since R3 has been sadly absent since Zen 2. It would cost too much both in design and manufacturing resource division between the two. Even then, the 4-core chiplet was basically a halved 8-core, so even if R3 magically returns with a 6-core offering, I don't see why it couldn't just be a halved 12-core chip.
Posted on Reply
#99
randomdude9182
LastDudeALiveThreadripper priced is probably an overstatement. A significant portion of Threadripper/Xeon's cost and value comes from the expanded platform. Quad/Octa channel memory, 80+ PCIe lanes, etc. Desktop won't have that. It will probably be ~$700+, but so will AMD's 24c/48t Zen 6 models, especially the X3D variants.
Pricing is not gonna be 700$ unless intel wants to bleed a rediculus amount of money, at minimum its $1100. Intel will be using 308 square mm of 2nm expensive sillicon, whereas amd is using 156mm squared of equally expensive 2nm sillicon. Intel would be bleeding rediculus amounts of money to make a 52 core nvl chip and price it so low. I suspect amd will also be like $750 for the 24 core 7ghz x3d model which might beat nvl anyways due to much higher clocks and not having to rely on E cores for the perf boost. They did say this will be a halfway house to HEDT (which costed ~$1500-$2000, and thier super high end chips (14900ks, ~$700 on launch). For it to be a halfway house, ill be interpreting it literally. That means the thing will be ~$1100 to ~1350 at worst, which makes sense as it would be impossible to price it near amd's top end by sheer die size alone).
Posted on Reply
Add your own comment
Nov 17th, 2026 13:51 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts