Question Zen 6 Speculation Thread

[MS_AT]

As far as X3D, that's something Dell's Alienware needs to get, because their "gaming desktop PCs" are a laughingstock.

That won't change even if they get x3D parts I am afraid.

 

[511]

But I was talking about Zen 6 SKUs. If the new AMD desktop IOD provide the features currently missing in Zen 4/5 desktop IODs, then Dell will be able to offer full AMD desktop as alternative to Arrow Lake / Nova Lake.

If you compare the platform of Z6 vs NVL. Z6 is clearly inferior
> No Integrated TB5
> 4 less Pci-E gen 5 lanes
> BT/WiFi

 

[fastandfurious6]

View attachment 130762

🤣🤣🤣🤣🤣🤣🤣

 

[Joe NYC]

If you compare the platform of Z6 vs NVL. Z6 is clearly inferior
> No Integrated TB5
> 4 less Pci-E gen 5 lanes
> BT/WiFi

I don't think these move the needle by more than 1 degree.

And Dell is not going to discontinue selling Intel based computers. Just provide more alternatives.

 

[511]

I don't think these move the needle by more than 1 degree.

Depends tbh Wifi/BT Combo is something no one would say no to every pre built desktop has this.

 

[Thibsie]

TB (3 or 4 or 5) would be nice but since USB4 is essentially a TB3 with a couple things removed, it's not that bad. As long as USB4 is indeed included, if possible with DP capability.

Add-in boards with TB are overpriced like big time and often with a very narrow compatibility.

The rest is meh. I much prefer a proper Ethernet Intel controller (which means at least 2.5G) than a WiFi one (even Intel one).

 

[511]

TB (3 or 4 or 5) would be nice but since USB4 is essentially a TB3 with a couple things removed, it's not that bad. As long as USB4 is indeed included, if possible with DP capability.

Well NVL has TB5 along with retimers integrate into SoC.

 

[Joe NYC]

Depends tbh Wifi/BT Combo is something no one would say no to every pre built desktop has this.

Why do you think this will be a differentiating feature? Why would Zen 6 desktop not have this? It's a standard feature.

 

[Doug S]

TB (3 or 4 or 5) would be nice but since USB4 is essentially a TB3 with a couple things removed, it's not that bad. As long as USB4 is indeed included, if possible with DP capability.

Add-in boards with TB are overpriced like big time and often with a very narrow compatibility.

The rest is meh. I much prefer a proper Ethernet Intel controller (which means at least 2.5G) than a WiFi one (even Intel one).

The market is trending towards people preferring to use wifi more and more, even on desktops. Building in a wifi controller OEMs are going to want as a feature in 100% of builds makes sense. Building in an ethernet controller that OEMs are increasingly making a premium feature (especially in mobile, but I wouldn't be surprised to start seeing lower end desktops without it) does not.

 

[Joe NYC]

The market is trending towards people preferring to use wifi more and more, even on desktops. Building in a wifi controller OEMs are going to want as a feature in 100% of builds makes sense. Building in an ethernet controller that OEMs are increasingly making a premium feature (especially in mobile, but I wouldn't be surprised to start seeing lower end desktops without it) does not.

Who is having problems with WIFI in 2025? It seems like one of the technologies that just works and has worked for more than a decade.

I don't think there is a single OEM PC sold without built in WIFI. Same with Bluetooth...

This seems like a conversation from 2020 or 2015 or 2010 or even 2005.

 

[Abwx]

Who is having problems with WIFI in 2025? It seems like one of the technologies that just works and has worked for more than a decade.

Only the digital part of wifi and Bt, wich is negligible, can be integrated,
the biggest part, wich is the RF circuitry, will keep being external.

Integrating the digital part complicate things, that s just a mean to grab a few $
at the expense of MB s manufacturing versatility.

 

[Joe NYC]

Only the digital part of wifi and Bt, wich is negligible, can be integrated,
the biggest part, wich is the RF circuitry, will keep being external.

Integrating the digital part complicate things, that s just a mean to grab a few $
at the expense of MB s manufacturing versatility.

And no difference to the end user, or something to take into account for purchasing decisions.

 

[Abwx]

And no difference to the end user, or something to take into account for purchasing decisions.

For the average joe it will change nothing, for whom is tech oriented it means
that on laptops wifi parts wont be upgradable since the current connectors
will be suppressed, that s a serious regression.

 

[dr1337]

integrating wifi/bt into x86 just seems like a massive waste of transistors

might make sense in an intel type chip where the IP can be its own die on a much cheaper node, but I really doubt we see it on anything AMD does. Especially for mobile SOCs where die space is critically important.

 

[adroc_thurston]

integrating wifi/bt into x86 just seems like a massive waste of transistors

Intel does this with their PCHs and PCDs?
That's why CNVio exists.

 

[511]

Who is having problems with WIFI in 2025? It seems like one of the technologies that just works and has worked for more than a decade.

I don't think there is a single OEM PC sold without built in WIFI. Same with Bluetooth...

This seems like a conversation from 2020 or 2015 or 2010 or even 2005.

Ironically AM5 users who wants to put Intel Wifi 7 solution in their MB.

 

[Joe NYC]

Ironically AM5 users who wants to put Intel Wifi 7 solution in their MB.

Intel networking products have been good. But it's something only nerds care about, because these are now commodities in client space.

I spent 2-3 hours going over AM5 mobo specs when I was picking a motherboard, but by now, I forgot what I have for networking or WIFI, because it makes zero difference.

 

[Abwx]

Ironically AM5 users who wants to put Intel Wifi 7 solution in their MB.

We are back to the old intel marketing with the CPU accelerating internet connections,
i mean, anything above wifi 4 is just a gimmick, wifi 4 is up to 600 Mb/s.

 

[511]

i mean, anything above wifi 4 is just a gimmick, wifi 4 is up to 600 Mb/s.

Only if it ships with 5 Ghz band 😭😭

 

[Doug S]

Only the digital part of wifi and Bt, wich is negligible, can be integrated,
the biggest part, wich is the RF circuitry, will keep being external.

Integrating the digital part complicate things, that s just a mean to grab a few $
at the expense of MB s manufacturing versatility.

Once you're doing chiplets there is nothing stopping you from putting the RF "on chip". Maybe it complicates things signal wise enough you don't want to do that, but you already need components on a board to support a CPU, that's just a special case (and I suppose you could skip it if you were doing something like an ATM where you specificially don't want to support wifi/BT)

 

[adroc_thurston]

Once you're doing chiplets there is nothing stopping you from putting the RF "on chip"

You want analog shoreline somewhere offdie and on a cheap node and expensive stuff (DSP cores and MACs) on cutting edge.
Which is exactly what Intel's been doing for a while.

 

[Abwx]

Once you're doing chiplets there is nothing stopping you from putting the RF "on chip". Maybe it complicates things signal wise enough you don't want to do that, but you already need components on a board to support a CPU, that's just a special case (and I suppose you could skip it if you were doing something like an ATM where you specificially don't want to support wifi/BT)

A wifi card is basicaly the same as a phone LTE modem, good luck integrating such a huge part in a chiplet, beside it would result in a catastrophic signal/noise ratio for the receiver, for obvious reasons the RF circuitry should be kept as far as possible from high frequencies sources, and the CPU is a huge one.

Also how would you get from the chiplet to the antennas connectors.?..
At such a distance it would require a piece of coaxial cable as it s impossible
to implement such a transmission line on the MB, and even if it was this would collect all
GHz parasistic signals that get out of the CPU, not counting harmonics from the PLL
if it s external as well as the ones from any high frequency switching circuits like USB parts.

 

[OneEng2]

And come on, if N2P CCDs are smaller and can clock 7-10% higher, that absolutely allows for higher prices and margins, they'd probably leave more money on the table by using N3P.
That's a non-argument.

No, it doesn't. HEDT and gaming markets have a price point that is driven by the market, not by the performance within that segment. The highest end gaming CPU (as an example) at launch (it's highest price point in its life cycle), is priced at $500-$700. It doesn't matter how it performs. The best performing will likely be at the top end of this range while the lower performing option (by OEM) will be near the bottom. Going from one generation to the next does not raise the market price (although inflation eventually does to a degree).

Nevermind that one issue you're ignoring is that with 12 cores and 48MB L3, N3P might simply not be dense enough to keep the CCD size compatible with the AM5 package, especially for the 2xCCD models, so the N2 logic transistor shrink might even be necessary in that regard, with the perf/efficiency as additional argument on top.

Possibly. I haven't done the math.

b) AMD isn't getting into OEM DT any time soon

Your mouth straight to God's ear. It must be true.

Intel does not seem to do very well when CPU and memory controllers are not on the same die. We will see if NVL finally cracks this problem for Intel.

Agree. It is for this reason that I hold out hope that NVL will actually post meaningful performance improvements.

A16 should be there.

Yes, but A16 may not offer enough of a benefit over N2 to justify an entire family design change. Moving forward, it may require a double die shrink to make it worthwhile to move to a new process, and thus a new generation of processors.

Processors are only redesigned when transistor budget increases provide a path to meaningful improvements to the product. When a single die shrink offered double the transistor budget, this was a no brainer. When a single die shrink offers 10-15% higher transistor density ..... and costs 30-50% more? Not so much.

I feel many people here overlook (or don't agree) that this is true and that from one generation to another, just using the same transistor budget, but doing it more cleverly, will give you very meaningful gains. I just don't believe this is true. I think it requires more transistors to give meaningful gains in processing.

 

[adroc_thurston]

It doesn't matter how it performs

yeah it does which is why ARL-S is toilet cheap.

 

[branch_suggestion]

No, it doesn't. HEDT and gaming markets have a price point that is driven by the market, not by the performance within that segment. The highest end gaming CPU (as an example) at launch (it's highest price point in its life cycle), is priced at $500-$700. It doesn't matter how it performs. The best performing will likely be at the top end of this range while the lower performing option (by OEM) will be near the bottom. Going from one generation to the next does not raise the market price (although inflation eventually does to a degree).

What matters is the ASP of the main margin additive part.
For Z4 it was $450, Z5 $480 and Z6 it's gonna be $600.
That more than overcomes the BOM increase, N2P is the right call and volumes will be similar in time, which means more revenue and more GM than prior generations.
This is exactly what NV has done for each generation since Maxwell with a short break for Ampere (node was dirt cheap so GMs were still very good), it works when you win and can only be stopped when you lose.

And to hell with it, NVL loses to Glymur let alone Z6, Intel is a banana republic.