• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

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

Page 550 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I don't see how AMD sets the MSRP for the 9700X at $299 when the MSRP for the 7700X was $399. It is true though that Zen 4 prices cratered pretty quickly. Maybe AMD learned from that. I doubt it. They can be their own worst enemy at times.

Oops, I think I got myself confused a little bit here (with pricing)...
 
Based on what? Overblown expectations?
Based previous performance uplift from zen 2, 3 and 4.
We need to keep in mind that datacenter is a more strategic area for AMD, and in datacenter (and HPC), there are more uses of AVX-512. For example, AI differencing running on CPU, without a GPU. That may be far more economical approach for lightly loaded system without spending arm and a leg on NVidia hardware.

We may hear more about this at the time of official Turin release.
I did say "for us plebs". Still waiting for an official review but it doesn't seem it will be anything different than what Amd said aka 16% is 16% and not anything else.
 
It’s funny when people call Zen 5 performance mid.

I do wonder what they would called the Skylake -> Kabylake increase. Junk? (tbh, it was)

Seriously we live in the best times in the CPU industry. OEMs using ARM CPUs as measurements, Intel bringing its A game again and it’s a strong contrast to the GPU world where everything is Nvidia dominated.
 
Based previous performance uplift from zen 2, 3 and 4.

I did say "for us plebs". Still waiting for an official review but it doesn't seem it will be anything different than what Amd said aka 16% is 16% and not anything else.
Well, there are multiple things at play.

  • Node jumps are getting smaller and smaller, meaning you can cram less additional circuitry with each generation. So IPC gains are getting harder to come by. Still, 16% is the second largest gen-on-gen jump;
  • Clock speed gains were bound to run out at some point as well;
  • Third, N3B being sort of a dud and Zen 5 being back ported to N4 means that some stuff was likely cut. I find it very curious that AMD were hell-bent on keeping the CCD area the same (under 70mm^2). That leads me to think the design team had limited transistor budget, so it makes sense that the jump isn't as big as some hoped.
Don't get me wrong, I myself wished for low 20s IPC jump. But 16% is still pretty respectable, albeit not earth-shattering.
 
I did say "for us plebs". Still waiting for an official review but it doesn't seem it will be anything different than what Amd said aka 16% is 16% and not anything else.

Yeah. Assuming that there was a certain budget for die area, and AVX-512 spend a good chunk of that budget, there is not much left "for us plebs" in other areas.

Zen 4 and Zen 5 are still short of critical mass for game developers to seriously consider AVX-512 optimizations. It will probably have to wait until next generations of Consoles come out with AVX-512.

The transition gamers from using Intel processors or AMD AM4 processors -> AM5 processors (which have AVX-512) would also need to kick into higher gear...
 
Ugh, no, logic is the only thing that scales.
What you can't pile on is cache or shoreline I/O.

I think the way to min-max this is to have 2 dies stacked on top of each other, one N6, the other in advanced node, and use Wafer on Wafer stacking, so that the default processor is a stacked processor, that is highly cost (plus power and performance optimized)
 
Last edited by a moderator:
Back
Top