• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Question Zen 6 Speculation Thread

Page 101 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
GTracing said:
AMD did announce that Zen6 has already taped out. 2026 isn't 100% confirmed, but it is very likely at this point. The chip would have to require some major rework to slip to 2027.
Click to expand...
Not to mention it would be an abysmal track record considering they initially tried to hit 12-15 months for new generations. Currently it has been
  • Zen 2 -> Zen 3 -> 16 months
  • Zen 3 -> Zen 4 -> 22 months
  • Zen 4 -> Zen 5 -> 22 months
  • Zen 5 -> Zen 6 (in 2027) -> 29+ months

I wouldn't hesitate to call it "an epic fail in delivery schedule" considering Zen 6 is supposed to be a rather minor architectural change.
 
AMDs "tick" = core arch changes & sometimes proc node shrink

AMDs "tock"= uncore arch changes & sometimes proc node shrink.



Sometimes they "tick-tock" in a single gen (Zen 2, Zen 4).
 
Last edited:
Abwx said:
They are tied by TSMCs yields, why release anything early if half the dies are not functional.?
Click to expand...

Because N2 yield is looking good, defect density D0 already <0.2:
www.computerbase.de

TSMC-Fortschritte: N2 mit Defektrate wie N3, N5 und N7, großer Erfolg erwartet

Viele Gerüchte rankten zuletzt um TSMCs Ausbeute bei neuen Technologiestufen. Zum Technology Symposium 2025 gibt es Antworten.
www.computerbase.de www.computerbase.de
1-2160.096ad06f.png
 
Intel reported a D0 of <0.4 back in September 2024. Which is pretty good and similar to what TSMC is reporting, when Intel wants to release 18A products in H2/2025. The question is, if 18A has the same D0 reduction trend line.
 
basix said:
Because N2 yield is looking good, defect density D0 already <0.2:
www.computerbase.de

TSMC-Fortschritte: N2 mit Defektrate wie N3, N5 und N7, großer Erfolg erwartet

Viele Gerüchte rankten zuletzt um TSMCs Ausbeute bei neuen Technologiestufen. Zum Technology Symposium 2025 gibt es Antworten.
www.computerbase.de www.computerbase.de
1-2160.096ad06f.png
Click to expand...
That s still too high, AMD never used the other processes at this point of the curves, rather when it reach +2Q, wich amount to one year from now plus the necessary time to have enough inventory, so a total of roughly 6 quarters, wich lead us to early Q4 2026 at best.
 
Gideon said:
Not to mention it would be an abysmal track record considering they initially tried to hit 12-15 months for new generations. Currently it has been
  • Zen 2 -> Zen 3 -> 16 months
  • Zen 3 -> Zen 4 -> 22 months
  • Zen 4 -> Zen 5 -> 22 months
  • Zen 5 -> Zen 6 (in 2027) -> 29+ months

I wouldn't hesitate to call it "an epic fail in delivery schedule" considering Zen 6 is supposed to be a rather minor architectural change.
Click to expand...
Blame Intel for AMD’s slow cadence. AMD doesn’t really have any competition, so they aren’t going to speed run advancements.
basix said:
Intel reported a D0 of <0.4 back in September 2024. Which is pretty good and similar to what TSMC is reporting, when Intel wants to release 18A products in H2/2025. The question is, if 18A has the same D0 reduction trend line.
Click to expand...
Intel doesn’t have bad yields for the same reason TSMC doesn’t: EUV. EUV does much of the heavy lifting, so defect rates will be similar, especially since ASML is the only game in town.

That is why I’ve always called out the skeptical folks that claim yields are an issue. They aren’t and with all their recent stuff using EUV, never were.

For Intel: capacity is an issue, cost is an issue, supply chains are an issue, and brain drain is an issue.

Regarding capacity, while Intel does have capacity, they don’t have enough to support multiple nodes and multiple market segments (OEMs, DIY, server, etc) They also need some for outside customers, so it has been a pain on their end.

Intel ships a TON of stuff. Huge volumes of chips. More than AMD. Shoot, their volume at times has exceeded what TSMC was capable of putting out. That will probably happen again with TSMC N2*. We will see, however.
 
Abwx said:
That s still too high, AMD never used the other processes at this point of the curves, rather when it reach +2Q, wich amount to one year from now plus the necessary time to have enough inventory, so a total of roughly 6 quarters, wich lead us to early Q4 2026 at best.
Click to expand...
It is high when manufacturing now. But still >80% yield for a small <70mm2 Die. Would be acceptable.

And in a year from now you absolutely can produce at good yields. The difference between +1Q and +2Q is not too big if talking in absolute yields and you could easily accept slightly worse yields for the sake of time to market.

But anyways: My point is, that N2 yield is not a problem for launching mid 2026. Tapeout has been done already as well. So if Zen 6 releases by end of 2026 or even 2027, something went wrong.
 
jpiniero said:
There's no need to rush it out if it's not faster than the 9800X3D in games.
Click to expand...
It is not about rushing. It is about market share and ASP. The earlier AMD can release Zen 6, the more market share they get. And the could charge higher prices for it.

Nova Lake supposedly is a 2026 product as well. If you are first on the market, you definitely have some competitive edge.

And my personal opinion: 8C is OK, but not "good" anymore. 12C sounds much nicer in 2026.

And there are servers and notebooks, where AMD would win even more with a power efficient CPU.
 
jpiniero said:
There's no need to rush it out if it's not faster than the 9800X3D in games.
Click to expand...
Zen 5 product release order was
  1. mobile and generic desktop,¹
  2. server,
  3. 1337 g4m1ng desktop.
I expect Zen 6 products to appear in similar order.² So yes, if asked I would not recommended to hold one's breath for the top-end video game toys segment to receive its Zen 6 update.

¹) Mobile used to trail same-generation desktop, but not anymore.
²) Actually a preview towards Zen 6 mobile has already been released by way of Strix Halo.
 
StefanR5R said:
Zen 5 product release order was
  1. mobile and generic desktop,¹
  2. server,
  3. 1337 g4m1ng desktop.
I expect Zen 6 products to appear in similar order.² So yes, if asked I would not recommended to hold one's breath for the top-end video game toys segment to receive its Zen 6 update.

¹) Mobile used to trail same-generation desktop, but not anymore.
²) Actually a preview towards Zen 6 mobile has already been released by way of Strix Halo.
Click to expand...
They're not releasing in that order this time around.
 
Josh128 said:
It will undoubtedly be faster than the 9800X3D in games, at least the X3D version, that is.
Click to expand...
Besides of general clockrate and IPC increases: As the new IOD will likely enhance memory throughput and latencies, the L3$ increases by +50% and I speculate that the core gets better latency hiding capabilities on top (bigger INT register file, bigger reorder buffer, better branch predictor, maybe bigger L1I cache), I somehow get tempted to assume, that even the non-X3D SKUs could beat the 9800X3D in games. Zen 6 non-X3D needs to be roughly 1.3x faster than a 9700X to achieve this and I believe that is possible with aforementioned updates. Zen 4 vs. Zen 3 was already right up there (on average the 7700X is roughly on par with a 5800X3D).
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top