Discussion Beyond zen 6

[Joe NYC]

grok summary

• 10:49–12:52 — Grimlock Halo (premium laptop APU): Renders of the 36+ core "mega" variant, with ~20 cores on the IOD base (mix of full Zen 7 + dense Zen 7C cores, possibly extra low-power ones), plus two 8-core chiplets added via bridge dies for expansion. Looks superficially similar to Strix Halo (Zen 5), includes monolithic base with GPU compute units. Emphasizes high core counts for future premium laptops.

https://twitter.com/x/status/2027481614268436513

Seems like a downgrade, even to Medusa Halo, which has a large GPU chiplet.

It seems like AMD designs are regressing, by trying to avoid advanced packeging, at cost of ceding to Intel (or Intel + NVidia) all of the advanced / premium / interesting niches, only competing for Cinememe niche, it seems, doing even that poorly.

It may turn out that the Grimlock Halo monolithic chiplet is a superb chilplet, exactly what a big portion of the market needs and it will be a runaway success, like hitting a jackpot.

But more likely, Intel with modularity and flexibility will have more OEM friendly designs.

It seems it is all about cost, AMD thinking that saving single digits $ on packaging will be a game changer, even in highest end, premium market.

I am thinking that, because the Silverton small optional dies look completely a useless upgrade.

For the one use of the Silverton, low cost AM5 desktop, if cost (and cost of packaging) is really that important, then why not just use the Grimlock Point? Like AMD has done with Phoenix / Hawk Point and now is doing with Gorgon Point?

Seems like someone with ADHD is in charge of AMD mobile roadmap.

 

[OneEng2]

Seems like a downgrade, even to Medusa Halo, which has a large GPU chiplet.

It seems like AMD designs are regressing, by trying to avoid advanced packeging, at cost of ceding to Intel (or Intel + NVidia) all of the advanced / premium / interesting niches, only competing for Cinememe niche, it seems, doing even that poorly.

It may turn out that the Grimlock Halo monolithic chiplet is a superb chilplet, exactly what a big portion of the market needs and it will be a runaway success, like hitting a jackpot.

But more likely, Intel with modularity and flexibility will have more OEM friendly designs.

It seems it is all about cost, AMD thinking that saving single digits $ on packaging will be a game changer, even in highest end, premium market.

I am thinking that, because the Silverton small optional dies look completely a useless upgrade.

For the one use of the Silverton, low cost AM5 desktop, if cost (and cost of packaging) is really that important, then why not just use the Grimlock Point? Like AMD has done with Phoenix / Hawk Point and now is doing with Gorgon Point?

Seems like someone with ADHD is in charge of AMD mobile roadmap.

The big difference between Intel and AMD these days is that one of them is making money while the other one is burning it like fire wood.

 

[Joe NYC]

The big difference between Intel and AMD these days is that one of them is making money while the other one is burning it like fire wood.

Speaking of which, there was an investor conference today, with Lisa Su as a guest, was commenting on the state of business.

Broadly, within the CPU sphere, she said she is seeing demand for Venice like never before (everyone wants it from day 1). But as far as client vs. server:
- server growing strongly in TAM and AMD growing market share within it.
- client still good but she conceded H2 may be slower, influenced by memory prices etc.

As far as AMD vs. Intel, AMD is extremely strong in server, widening the gap with Zen 6.
Intel strong area is client.

The area where AMD is strong is growing
The area where Intel is strong is stagnant or could be shrinking.

 

[Geddagod]

Speaking of which, there was an investor conference today, with Lisa Su as a guest, was commenting on the state of business.

Broadly, within the CPU sphere, she said she is seeing demand for Venice like never before (everyone wants it from day 1). But as far as client vs. server:
- server growing strongly in TAM and AMD growing market share within it.
- client still good but she conceded H2 may be slower, influenced by memory prices etc.

As far as AMD vs. Intel, AMD is extremely strong in server, widening the gap with Zen 6.
Intel strong area is client.

The area where AMD is strong is growing
The area where Intel is strong is stagnant or could be shrinking.

Honestly, unless Zen 6 mobile blows everyone away with how good the uncore power draw is, Intel could be "growing" in client. Shipping volume can be impacted by many things, but at least in terms of how well they compete.
Simply because I refuse to believe things can get worse than ARL-S vs Zen 5X3D in desktop. No shot. Meaning that in desktop Intel may be able to regain a bit of share...
And mobile ofc Intel is still very strong, in large part due to the efforts of all the teams except the core teams lol. So I expect this to stagnate.

 

[adroc_thurston]

Simply because I refuse to believe things can get worse than ARL-S vs Zen 5X3D in desktop. No shot. Meaning that in desktop Intel may be able to regain a bit of share...

lmao

 

[Abwx]

.

 

[OneEng2]

Speaking of which, there was an investor conference today, with Lisa Su as a guest, was commenting on the state of business.

Broadly, within the CPU sphere, she said she is seeing demand for Venice like never before (everyone wants it from day 1). But as far as client vs. server:
- server growing strongly in TAM and AMD growing market share within it.
- client still good but she conceded H2 may be slower, influenced by memory prices etc.

As far as AMD vs. Intel, AMD is extremely strong in server, widening the gap with Zen 6.
Intel strong area is client.

The area where AMD is strong is growing
The area where Intel is strong is stagnant or could be shrinking.

I think AMD would be perfectly happy to utilize every 2nm wafer it produces to fill the highest profit market in x86 😉.

They also make pretty good coin on the HEDT and X3D gamers.

At some point, seems like AMD will need a plan to move into the more commodity and higher volume markets, but its hard to argue with their current business strategy.

 

[Joe NYC]

I think AMD would be perfectly happy to utilize every 2nm wafer it produces to fill the highest profit market in x86 😉.

They also make pretty good coin on the HEDT and X3D gamers.

At some point, seems like AMD will need a plan to move into the more commodity and higher volume markets, but its hard to argue with their current business strategy.

Everything AMD will be making on N2 will be extremely profitable. Fortunately, there is no competition for N2 wafers from NVidia, and other, consumer oriented customers (Apple, Qualcomm, MediaTek) of N2 will likely see tepid demand due to memory pricing.

So, AMD may have the door wide open to up their N2 orders.

As far as volume markets, say Hello to Gorgon Point, it will be with us for a long time, and also more Hawk Point.