Question Zen 6 Speculation Thread

[Doug S]

moooooooooooooom moooom mom get the camera i'm in the TV
View attachment 129562

And the author is so clueless he doesn't even understand that even if Q3 2026 is when they ramp that we won't see finished packaged chips until late Q1 2027 at the absolute soonest - and then only if AMD gets some of the first N2P wafers AND they use them for Zen 6 CPUs rather than something more profitable like server or AI/GPU.

 

[adroc_thurston]

And the author is so clueless he doesn't even understand that even if Q3 2026 is when they ramp

Oh no noo no ramp in TSM lingo means volume wafer churn.
Intel 'ramp' is HVM in TSM lingo.

 

[marees]

Nvidia responded exactly the same way they did back with Iris Pro - cutting the price of the low end parts so there's no cost advantage.

You get iGPUs for three reasons:
-Cost
-Battery life
-Form factor

The regular iGPUs achieve all three marvelously. The Halo iGPUs increase cost, and the smaller form factor generally further increases cost. In addition, it's pretty much useless in desktops because the price factor gets further compounded by having it all in a single chip.

Part of the fault is in the vendor that makes them. If Intel had Iris Pro priced at i3 + Iris Pro, then it would have been more competitive. Same with AMD. Why not offer Ryzen R3 level with the Strix Halo? Both vendors make Halo iGPUs with a fundamentally different reason than regular iGPUs - make more money. They pair halo iGPUs with halo CPUs to force upsell. Also, modern iGPUs are in reality not really "free" either. Because back in the chipset graphics days, you could pair the lowest CPU with the best chipset graphics. Now you have to spend $100 or more for the highest. Are you *really* saving money? It's all a clever marketing trick. Heck, if you really cared about consumer value, you'd offer Strix Halo/Iris Pro graphics with a Celeron CPU and price it accordingly.

Which Nvidia can attack easily since they are an established vendor.

Medusa premium fixes this.

It is the replacement for Z2E — effectively it is the Z3E

 

[Doug S]

Oh no noo no ramp in TSM lingo means volume wafer churn.
Intel 'ramp' is HVM in TSM lingo.

They haven't even STARTED mass production of N2 yet. You think there will be finished N2P wafers in Q3 next year - which would be six months or less after the first finished N2 wafers? Dream on!

 

[511]

They haven't even STARTED mass production of N2 yet. You think there will be finished N2P wafers in Q3 next year - which would be six months or less after the first finished N2 wafers? Dream on!

Let him dream N3P was in HVM last year and the products only launched like Q3 this year In fact MI355X is the first N3P product to be announced.
If N2P is HVM Next year no way you can get the product the same year at the same time.
TSMC is kind of at fault as well they just marked N2/N2P pdk as one and the same.

 

[ToTTenTranz]

Why not offer Ryzen R3 level with the Strix Halo? Both vendors make Halo iGPUs with a fundamentally different reason than regular iGPUs - make more money.

Because somehow AMD's board convinced themselves they could jump straight into Apple's ASP and margins per SoC without having even a fragment of Apple's brand perception and software ecosystem that they spent decades building upon.

I.e. AMD isn't pairing a full Halo IOD with e.g. a single CCD at a competitive price out of hubris and a failing market research team. Strix Halo isn't selling, but AMD is apparently willing to die on the "I'd rather not sell it at all if I can't have a >100% margin" hill.

And the fact that virtually no one picked up STX Halo for laptops means AMD apparently didn't even consult laptop makers to know what they'd like to see and how much they'd be willing to pay.


Hopefully, @adroc_thurston will be wrong / plans will change again and Medusa Premium + Halo will have a perceived value that put them on laptop+handheld makers' roadmaps instead of driving them away like they're doing with Strix Halo.

 

[marees]

Because somehow AMD's board convinced themselves they could jump straight into Apple's ASP and margins per SoC without having even a fragment of Apple's brand perception and software ecosystem that they spent decades building upon.

I.e. AMD isn't pairing a full Halo IOD with e.g. a single CCD at a competitive price out of hubris and a failing market research team. Strix Halo isn't selling, but AMD is apparently willing to die on the "I'd rather not sell it at all if I can't have a >100% margin" hill.

And the fact that virtually no one picked up STX Halo for laptops means AMD apparently didn't even consult laptop makers to know what they'd like to see and how much they'd be willing to pay.


Hopefully, @adroc_thurston will be wrong / plans will change again and Medusa Premium + Halo will have a perceived value that put them on laptop+handheld makers' roadmaps instead of driving them away like they're doing with Strix Halo.

I don't quite get your point

In general I see laptop OEMs promote CPU rather than igpu

That means medusa point & medusa point + extra CCD is what normal laptop makers are interested in

There is no series S next gen
So xbox will make a big push for medusa premium — it can slot into strix point / medusa point socket — as Z3E aka 'xbox' handheld apu, to be made by OEMs. This should sell for around $800 to $1000 just like the rog 'xbox' Ally X today

I don't expect medusa halo in any laptops/tablets/NUCs cheaper than $2000

 

[ToTTenTranz]

In general I see laptop OEMs promote CPU rather than igpu

That means medusa point & medusa point + extra CCD is what normal laptop makers are interested in

Why didn't those same laptop OEMs design and massively promote models with the 16-core 395 MAX?

There is no series S next gen
So xbox will make a big push for medusa premium — it can slot into strix point / medusa point socket — as Z3E aka 'xbox' handheld apu, to be made by OEMs. This should sell for around $800 to $1000 just like the rog 'xbox' Ally X today

There's some confusion here. Xbox's Magnus uses AT2, not AT3. Medusa Halo uses AT3 and Medusa Premium uses AT2.
The closest thing we'll ever get to a xbox handheld APU / Z3E will be the AT4 SoC, though its gaming performance will probably suck until it's paired with LPDDR6.

 

[marees]

Why didn't those same laptop OEMs design and massively promote models with the 16-core 395 MAX?





There's some confusion here. Xbox's Magnus uses AT2, not AT3. Medusa Halo uses AT3 and Medusa Premium uses AT2.
The closest thing we'll ever get to a xbox handheld APU / Z3E will be the AT4 SoC, though its gaming performance will probably suck until it's paired with LPDDR6.

There is no confusion

AT4 is in medusa premium = Z3E = 'xbox' handhelds (this uses exactly same socket as strix point & medusa point)

 

[marees]

Why didn't those same laptop OEMs design and massively promote models with the 16-core 395 MAX?

Strix halo chip size is much larger compared to medusa point with extra cpu ccd. So the costs are in different range

 

[marees]

What is Medusa Baby / Medusa BB

AMD mobile CPU roadmap leak claims Zen 6 arrives in 2027

As Zen 6 arrives to high-end, Zen 2 dominates low-end laptops.

www.tomshardware.com

 

[ToTTenTranz]

Strix halo chip size is much larger compared to medusa point with extra cpu ccd. So the costs are in different range

But AMD still launched Strix Halo for laptops and it landed in almost no design wins. Strix Halo in laptops ended up having somehow even worse popularity than AMD's RDNA3 dGPUs for mobile, which was quite the feat.
Either the price, the performance or the features were terribly communicated to laptop OEMs here. AMD made Strix Halo a product its clients didn't want.

There is no confusion

AT4 is in medusa premium = Z3E = 'xbox' handhelds (this uses exactly same socket as strix point & medusa point)

Ok I think I understood now. You're suggesting Microsoft will try to make AT4 to AT2 what Series S was to Series X, in terms of target specs for 1st party developers?

 

[marees]

You're suggesting Microsoft will try to make AT4 to AT2 what Series S was to Series X, in terms of target specs for 1st party developers?

Yes

 

[marees]

But AMD still launched Strix Halo for laptops and it landed in almost no design wins. Strix Halo in laptops ended up having somehow even worse popularity than AMD's RDNA3 dGPUs for mobile, which was quite the feat.
Either the price, the performance or the features were terribly communicated to laptop OEMs here. AMD made Strix Halo a product its clients didn't want.

You can think of strix halo / medusa halo with AT3 as the equivalent of Kaby lake G

How many laptop wins did kaby lake G get ??

 

[ToTTenTranz]

You can think of strix halo / medusa halo with AT3 as the equivalent of Kaby lake G

How many laptop wins did kaby lake G get ??

One.

And like AMD on Strix Halo, I don't think anyone at Intel was exactly happy with how Kaby Lake G ended.

Though the Vega M had a problem STX Halo's iGPU doesn't: a short amount of VRAM made up of very expensive HBM.

 

[DrMrLordX]

What is Medusa Baby / Medusa BB

Also why is Medusa Halo not on this roadmap?

 

[marees]

Also why is Medusa Halo not on this roadmap?

Either it is Not of interest to this laptop maker or medusa halo & premium are penciled in for CES 2028

 

[igor_kavinski]

One.

Two?

 

[511]

Two?

OG APU with HBM

 

[fastandfurious6]

View attachment 129539

taiwan numba one 🇹🇼 🇹🇼 🇹🇼 🇹🇼 you can see how wholesome proud they feel hahaha

 

[adroc_thurston]

One.

And like AMD on Strix Halo, I don't think anyone at Intel was exactly happy with how Kaby Lake G ended.

Though the Vega M had a problem STX Halo's iGPU doesn't: a short amount of VRAM made up of very expensive HBM.

Wrong, Kaby-G had 4 (or more, I forgot).
Had a x360 Spectre and an XPS.

 

[SteinFG]

What is Medusa Baby / Medusa BB

View attachment 129601

AMD mobile CPU roadmap leak claims Zen 6 arrives in 2027

As Zen 6 arrives to high-end, Zen 2 dominates low-end laptops.

www.tomshardware.com

MLID says it's either a 2+2 or a 2+4 zen 6/6c, plus 2 CU igpu, plus 50 tops npu. But realistically, it's too far away to say.

 

[perry mason]

You can think of strix halo / medusa halo with AT3 as the equivalent of Kaby lake G

How many laptop wins did kaby lake G get ??

I know there's a lot of context provided by AMD's business model, including their historic reticence to assist on laptop design, but ultimately a lot of Apple buyers may have been interested in picking up Strix Halo if, you know, it actually came in a premium 16-inch model from a major OEM.

Where's the Dell Pro Max version? Why did HP exclusively put it in a 14 inch case? Imagine if they rolled it out with proper AI branding and support.

Instead, AMD leaned in to the gamerz ecosystem when Strix Halo was rolled out. Not the right move when looking for margin.

 

[511]

MLID says it's either a 2+2 or a 2+4 zen 6/6c, plus 2 CU igpu, plus 50 tops npu. But realistically, it's too far away to say.

Bruh wtf 50 TOPS it's not a CPU but a NPU with the spec. Microsoft is gonna brand this as Buy a copilot PC and get a CPU/GPU Free

 

[SteinFG]

Bruh wtf 50 TOPS it's not a CPU but a NPU with the spec

Look at Ryzen ai 5 330, I think Medusa BB will be a direct zen6 replacement, because specs line up.