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

[DisEnchantment]

Speculate at will

 

[jdubs03]

All the negative talk, a win for the consumer.

Nah, what was negative was AMD setting prices too high in the first place. Why else are they already discounting them? They’re not selling well enough. The 9900X price at the current discount is what it should’ve been for MSRP.

 

[Hail The Brain Slug]

Already seeing price drops. Look at that 9900X. 10% discount lol.

AMD Ryzen 9 9950X drops to $623, 9900X now $449 in the US, with prices slashed up to 12% in Germany - VideoCardz.com

Ryzen 9000 gets first price cuts Not official AMD price cuts, but retailer price cuts. The AMD Ryzen 9000 series has arrived, with at least four SKUs already announced and launched. However, a few things need to be finalized before the full transition to the new platform can be declared...

videocardz.com

BH always had the 9900X for $449. I think they missed price jacking it to $499 right before release.

 

[GodisanAtheist]

So... when is AMD gonna rehire Jim Keller?

 

[Abwx]

So... when is AMD gonna rehire Jim Keller?

The current crews are good enough, it s a long time that he does no more do CPU design, the top enginers that were in charge of Zen 1 are these ones, i dont see Keller in the pic :

Amid challenges, chipmaker AMD sees a way forward

For more than two decades, Advanced Micro Devices Inc. has been the comeback kid of computer-related electronics. \n The chipmaker has worked itself into difficult places only to make Houdini-like esc…

eu.statesman.com

 

[jpiniero]

All the negative talk, a win for the consumer.

I think it's pretty clear that AMD should not have bothered with releasing it to DIY. Once they realized it wasn't going to beat the 7800X3D in games.

 

[PJVol]

i dont see Keller in the pic

Also, no Naffziger and Kosonocky, Singh is there though

 

[Markfw]

I think it's pretty clear that AMD should not have bothered with releasing it to DIY. Once they realized it wasn't going to beat the 7800X3D in games.

gaming is not the only use for DIY. Thats a very specialized opinion

 

[Abwx]

No Naffziger and Kosonocky, Singh is there though

Nafziger was in charge of the power management research and implementations for whatever future CPU, so he was not exactly related to the core logic design itself.

Now that you mentioned his name he made a remarkable paper when he joined AMD, under the lines he pointed that the soon to be designed BD was not adequate for its time, but he was new at AMD and infortunately he didnt manage to influence the decisions makers.

 

[jpiniero]

gaming is not the only use for DIY. Thats a very specialized opinion

It's the vast majority of sales. That's why sales are bad.

 

[Markfw]

It's the vast majority of sales. That's why sales are bad.

1) prices are pretty high considering the reviews
2) Reviews are bad for games due to the windows "bug" that is still an issue, and most people don't run linux.

But for those of use that don't game, its great !

 

[PJVol]

I think it's pretty clear that AMD should not have bothered with releasing it to DIY.

Wondering, if the decision to go 4N path for DT parts was made from the start or at some later stage?

 

[JustViewing]

I think it's pretty clear that AMD should not have bothered with releasing it to DIY. Once they realized it wasn't going to beat the 7800X3D in games.

There is more to a CPU than gaming alone. In addition, you need to have top tier graphics card to realize the gaming performance, specially top tier Nvidia graphics card because of its driver overhead.

 

[gdansk]

I think it's pretty clear that AMD should not have bothered with releasing it to DIY. Once they realized it wasn't going to beat the 7800X3D in games.

There's no problem releasing it for DIY. But for its utility they need to adjust their marketing and pricing.

 

[DrMrLordX]

So... when is AMD gonna rehire Jim Keller?

He's busy at Tenstorrent. So probably never. Keller will probably want to retire soon-ish.

 

[marees]

It's the vast majority of sales. That's why sales are bad.

If crypto mining takes off again that could boost zen 5 sales

 

[igor_kavinski]

Keller will probably want to retire soon-ish.

Is there a Keller Jr.?

 

[tsamolotoff]

2) Reviews are bad for games due to the windows "bug" that is still an issue, and most people don't run linux.

Looks like this 'bug' is that under true admin account the spectre/meltdown patches are either working in a different way, or not working at all. After I've disabled it, the difference between the benchmark results of FC6 and CP2077 disappeared. Could be something else, but I've rerun the tests several times and did several reboots to be sure.

 

[Cllaymenn]

🙂

 

[Abwx]

I wonder how many people worked on the ZEN5 architecture project, how many core designers? 🙂

I remember in an interview, I think Lisa was talking about how 2 teams worked on ZEN3 (and probably 4) independently of each other. Whose design turned out to be better, won and that architecture was introduced to the final version. I wonder if 2 teams worked on ZEN 5 in a similar way. 😀 But seeing how many problems ZEN 5 has, it's probably only one that didn't manage to finish the work for the much faster July premiere. 🙁

Dunno how much people are required, excluding Xilinx the total headcount is about 21 000, but there a lot of divisions for all possible products from servers to DT, APUs, networking, gaming and AI GPUs and of course general management and maketing, so eventually several thousands for the cores designs.

 

[igor_kavinski]

Dunno how much people are required

Rising Woman of Influence: Suzanne Plummer - Design The Solution

2019 Nominee Suzanne Plummer Corporate Vice President of Design Engineering AMD Suzanne Plummer is a corporate vice president of Design Engineering at AMD. In this role, she is responsible for […]

designthesolution.org

It was around 800 in 2019. I wonder how many managers report to her. It's possible that she was tasked with overseeing some of the AI crap and that affected her ability to look after the core CPU design teams.

 

[Abwx]

Rising Woman of Influence: Suzanne Plummer - Design The Solution

2019 Nominee Suzanne Plummer Corporate Vice President of Design Engineering AMD Suzanne Plummer is a corporate vice president of Design Engineering at AMD. In this role, she is responsible for […]

designthesolution.org

It was around 800 in 2019. I wonder how many managers report to her. It's possible that she was tasked with overseeing some of the AI crap and that affected her ability to look after the core CPU design teams.

Lisa Su own specialty is semiconductors design, not CPU design wich is surely delegated to Mark Papermaster with Mike Clark just below him.
Ironically Lisa Su s competences would had more suited as head of TSMC.

 

[Jan Olšan]

I think it's pretty clear that AMD should not have bothered with releasing it to DIY. Once they realized it wasn't going to beat the 7800X3D in games.

The same people that are 5%talking now would be mad that AMD is keeping the good stuff from them, also company would be called retarded by limiting it to OEMs etc.

 

[StefanR5R]

https://twitter.com/x/status/1825859218802618610

View attachment 105848

So the higher power consumption of Zen 5 carries to EPYC.

Higher power consumption compared to what?

Zen 4. Its already established that low clock / TDP Zen 5 is scarcely better than Zen 4 at all in perf/w (7700 vs 9700X @65W), and actually can be worse at low enough TDP.

Oh, you were referring to the sketchy scaling of GNR (and STX I think) to low per-core power levels. Yes, it will be interesting to see how Turin will manage. At the top end, there is a core count increase, the IOD will support faster RAM, the top end has more fabric links active... and at the same time the cores are that much wider. Quite some places to spend additional socket power at. Potentially one out of several debatable points why Turin gets a newer CCD stepping.

I was just re-reading David Huang's Strix Point analysis part 2 for unrelated reasons. It contains plots of SPECrate®2017 Integer rate1 performance over core power and over package power, for Strix Point and Phoenix. These plots show uniformly better performance of Strix Point at all power levels.

So, what "is already established" is in fact that the Zen 5 core is consistently better than the Zen 4 core at low power levels in integer workloads (and at high power levels too).

If default TDP of 96c Turin vs. 96c Genoa is raised as stated in this rumor, then higher supported memory clock alone would already be a sufficient reason to do this.

 

[Kocicak]

Outside of gaming, AMD has the best overall CPU with Zen 5, even on Windows (and especially on Linux).

No. Just no.

For gaming AMD 7800X3D is the best, it has no competition beside other more expensive and for gaming unnecessary AMD 3D CPUs.

For everything else Intel 13600/14600 CPUs are generally regarded to be the best universal CPUs with excellent price to performance ratio. 9 out of 10 AMD enthusiasts will confirm that.

 

[Cllaymenn]

🙂