Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

[DisEnchantment]

Speculate at will

 

[poke01]

Nice, jumping straight to number 300 so they can change the numbering scheme again in just 5 years.

Wonder how many wheels will be created in 5 years

 

[Gideon]

The next naming scheme is to rename all mobile SKUs: "rAIzen" with the tier being annotated by the amount of AIs in the row.

So "Ryzen R3" will become "RAIzen AI" but Ryzen R9 will be "RAIzen AI-AI-AI-AI!"

 

[SteinFG]

Now that the naming scheme is leaked, I started thinking - why are they reserving _70 and _65 for Ryzen 9? Probably has to do with Strix Halo and Fire range
16C Strix halo / Fire range: 390 (ryzen 9)
12C Strix halo / Fire range: 380 (ryzen 9)
4+8 Strix point: 370 (ryzen 9)
3+7 or 4+6 Strix point: 360 (ryzen 9)
4+4 Kraken: 350 (ryzen 7)
3+3 or 2+4 Kraken: 340 (ryzen 5)

 

[tsamolotoff]

So Ryzen AI 9 HX 370 and Ryzen AI 9 365 are allegedly the updated names for Strix Point, and of course still no leaked serious benchmarks 10 days before some disclosure...

AMD Ryzen AI 9 HX 370: Strix Point tritt mit Namen im Intel-Core-Ultra-Stil plus AI an

Die Drehscheibe war gestern, im Notebook setzt AMD mit Strix Point schon wieder auf eine neues Schema: Mit AI und Core-Ultra-Anleihen.

www.computerbase.de

Canonically, 365 should be based off a new version of the same architecture, while 370 should be a rebadge of a previously successful product. I wonder how and why do they do such rebrands, it's already very confusing with their wheels of doom and whatnot

 

[SteinFG]

Canonically, 365 should be based off a new version of the same architecture, while 370 should be a rebadge of a previously successful product. I wonder how and why do they do such rebrands, it's already very confusing with their wheels of doom and whatnot

There's no canon lol

 

[Kepler_L2]

Canonically, 365 should be based off a new version of the same architecture, while 370 should be a rebadge of a previously successful product. I wonder how and why do they do such rebrands, it's already very confusing with their wheels of doom and whatnot

There is no logic or reason, AMD marketing just makes stuff up as they go.

 

[Abwx]

Yeah the reasoing was particularily odd (from the same computerbase article, translated):

The kings of the trolls suggested something to AMD to be sure to be ahead in numbers, and it s right that you cant beat WCCF Tech when it comes to troll your way in the business...

AMD Doesn't Want A Numerical Disadvantage Against Intel's Core Ultra, Shifts To "Ryzen AI 300" Branding For Strix APUs

It looks like AMD has decided to drop the "Ryzen AI 100" branding for its Strix APUs and instead use "Ryzen AI 300" naming convention.

wccftech.com

https://cdn.wccftech.com/wp-content/uploads/2024/05/AMD-Strix-Point-Ryzen-AI-HX-APUs.jpg

 

[carancho]

With the increased competition from ARM CPU vendors we are going to get a hyperinflation in model numbers.

The best numbering scheme is Samsung's S line. Year + tier. Very simple, and if you want more information just look up benchmarks.

 

[adroc_thurston]

With the increased competition from ARM CPU vendors

?

 

[gdansk]

?

Presumably SDX.

 

[Kolifloro]

Some of you previously said ... " 'Strix Halo' will be pretty expensive ... " ... ... OK, then ... ... Apple's M4 will also be pretty expensive, but built over N3E ... since 'Halo' is going to be such expensive it would have been great to be manufactured on N3E node too ...

Yes, but time to market is key, and N3E just came out (and only on ipad lol). If N3E was delayd for some reason, Strix Halo would be delayed too.

While this argument has its merit (it really does) ... someone might argue that the context is 'universal' ... so if Apple is able to surf this wave ... AMD should too.

I admit that I tend to consider that the real reason may be related to a strategic business decision: perhaps AMD HQ thinks that Apple users are more likely to invest more money in laptops ... than AMD users. So competing with Apple for TSMC's top-notch manufacturing node chips could backfire ... because there might be less 'luxury' demand for AMD than expected ...

It would be a pity ... AMD has already overtaken Intel ... and it would be a really interesting challenge to see if AMD is also capable of surpassing Apple ... in the field of energy efficiency; and in the field of efficiency, the manufacturing node is a key factor.


Regards. 🙂



Ps.- The difference might be Zen 6 on 3nm ... or on 2nm ...

 

[Goop_reformed]

So Ryzen AI 9 HX 370 and Ryzen AI 9 365 are allegedly the updated names for Strix Point, and of course still no leaked serious benchmarks 10 days before some disclosure...

AMD Ryzen AI 9 HX 370: Strix Point tritt mit Namen im Intel-Core-Ultra-Stil plus AI an

Die Drehscheibe war gestern, im Notebook setzt AMD mit Strix Point schon wieder auf eine neues Schema: Mit AI und Core-Ultra-Anleihen.

www.computerbase.de

Yeah really weird. The weeks leading the radeon VII launch was very similar.

 

[Thibsie]

It's still better than whatever the USB Implementer Forum is doing...

ah, those idiots... Always renaming old USB to the new one just for the sake of $ and confusion.

 

[FlameTail]

It would be a pity ... AMD has already overtaken Intel ... and it would be a really interesting challenge to see if AMD is also capable of surpassing Apple ... in the field of energy efficiency; and in the field of efficiency, the manufacturing node is a key factor

The microarchitecture is more important than the node.

 

[FlameTail]

Kraken Point, Strix Point, Strix Halo

All three Zen5 mobile parts, from the lowest end to the highest end, will have a ~50 TOPS NPU. This is interesting. The NPU does not scale up in size/performance for the higher end parts, like CPU/GPU does. Why not?

Will it remain this way for future generations too?

 

[biostud]

Kraken Point, Strix Point, Strix Halo

All three Zen5 mobile parts, from the lowest end to the highest end, will have a ~50 TOPS NPU. This is interesting. The NPU does not scale up in size/performance for the higher end parts, like CPU/GPU does. Why not?

Will it remain this way for future generations too?

Because it is a check mark feature, that will have little impact on sales.

 

[Kepler_L2]

Kraken Point, Strix Point, Strix Halo

All three Zen5 mobile parts, from the lowest end to the highest end, will have a ~50 TOPS NPU. This is interesting. The NPU does not scale up in size/performance for the higher end parts, like CPU/GPU does. Why not?

Will it remain this way for future generations too?

Microsoft requirement

 

[Mahboi]

And beyond the fact that it's just an MS requirement, AMD's interest would rather be to play the same card as NV: "you want REAL AI? Buy a Radeon card".

 

[lightmanek]

And beyond the fact that it's just an MS requirement, AMD's interest would rather be to play the same card as NV: "you want REAL AI? Buy a Radeon card".

Or use your brian (the original I) 😀

 

[Timorous]

Or use your brian (the original I) 😀

Have you looked at the world recently?

 

[Mahboi]

I'm from the doomer EDM era, not the Metal era. I only know that Megadeth song from GTA Vice City.

 

[FlameTail]

We need quality memes like this, for Zen5.

 

[Glo.]

The links are broken on AT forums.

I clicked "Zen 5 speculation" thread on Anadtech Forums, and suddenly I found myself in the music thread.

😛

 

[Joe NYC]

Kraken Point, Strix Point, Strix Halo

All three Zen5 mobile parts, from the lowest end to the highest end, will have a ~50 TOPS NPU. This is interesting. The NPU does not scale up in size/performance for the higher end parts, like CPU/GPU does. Why not?

Will it remain this way for future generations too?

If NPU is a tax, you pay the minimum necessary tax.

 

[biostud]

If NPU is a tax, you pay the minimum necessary tax.

Once they realise that people are not willing to pay $20/month for co-pilot plus and similar AI services I wonder how important they will be.