Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)

Page 275 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Vattila

Senior member
Oct 22, 2004
821
1,457
136
Except for the details about the improvements in the microarchitecture, we now know pretty well what to expect with Zen 3.

The leaked presentation by AMD Senior Manager Martin Hilgeman shows that EPYC 3 "Milan" will, as promised and expected, reuse the current platform (SP3), and the system architecture and packaging looks to be the same, with the same 9-die chiplet design and the same maximum core and thread-count (no SMT-4, contrary to rumour). The biggest change revealed so far is the enlargement of the compute complex from 4 cores to 8 cores, all sharing a larger L3 cache ("32+ MB", likely to double to 64 MB, I think).

Hilgeman's slides did also show that EPYC 4 "Genoa" is in the definition phase (or was at the time of the presentation in September, at least), and will come with a new platform (SP5), with new memory support (likely DDR5).

Untitled2.png


What else do you think we will see with Zen 4? PCI-Express 5 support? Increased core-count? 4-way SMT? New packaging (interposer, 2.5D, 3D)? Integrated memory on package (HBM)?

Vote in the poll and share your thoughts! :)
 
Last edited:
  • Like
Reactions: richardllewis_01

Exist50

Platinum Member
Aug 18, 2016
2,452
3,106
136
And best of all is that Zen 5 will be the first one that fully utilize Xilinx AI hardware accelerators or something like that, and it is clear that AMD saved all resources and IPC gain for Zen 5, Intel is in real trouble if they can't beat Zen 4 that is only evolution of Zen 3, while Zen 5 will be revolution, complete redesign architecture with Xilinx product.
Well let's not get too ahead of ourselves. It looks like AMD's first integrating the Xilinx IP as an accelerator, not part of the core. And while Zen 5 will probably be a bigger architectural change than Zen 4, we've got a pretty good sense of their cadence. Wouldn't make sense for them to sandbag anyway.

Edit: Also, relative to Zen 4, they're not going to get nearly as big an improvement on the process side. Will be interesting to see how that manifests in the product.
 
Last edited:

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
Well let's not get too ahead of ourselves. It looks like AMD's first integrating the Xilinx IP as an accelerator, not part of the core. And while Zen 5 will probably be a bigger architectural change than Zen 4, we've got a pretty good sense of their cadence. Wouldn't make sense for them to sandbag anyway.

Edit: Also, relative to Zen 4, they're not going to get nearly as big an improvement on the process side. Will be interesting to see how that manifests in the product.
Would it matter if the Xilinx IP is at the Core or at the IO? They are shooting for an HPC/AI APU Revolution with CPUs, GPU and Xilinx IP on a single Package.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,106
136
Would it matter if the Xilinx IP is at the Core or at the IO? They are shooting for an HPC/AI APU Revolution with CPUs, GPU and Xilinx IP on a single Package.
Why assume they're trying to do all of that at once, especially for HPC/AI? For client, they'll want a relatively small AI accelerator, and for server, it either needs to be central to the product, or it doesn't make much sense to have at all. I think something like the MI300 with the GPU die swapped for AI would be a better config for that market.
 
  • Love
  • Like
Reactions: Tlh97 and Thibsie

turtile

Senior member
Aug 19, 2014
633
315
136
And best of all is that Zen 5 will be the first one that fully utilize Xilinx AI hardware accelerators or something like that, and it is clear that AMD saved all resources and IPC gain for Zen 5, Intel is in real trouble if they can't beat Zen 4 that is only evolution of Zen 3, while Zen 5 will be revolution, complete redesign architecture with Xilinx product.

Everyone is concentrating on IPC but they are forgetting that Zen 4 looks to be 35%+ faster at the same power for multicore. Zen 5 on 4nm isn't going to push clocks much so they would need to increase IPC by at least 30% or more to match the improvement from Zen 3 to 4. I assume AMD will use the 4nm to make the core slightly bigger, then shrink it to 3nm for the massive 200+ core part for the cloud.
 

Thunder 57

Diamond Member
Aug 19, 2007
4,027
6,741
136
That's part of it, but didn't Userbenchmark have some kind AMD review that read as if some angsty teenager wrote it? Maybe I'm misremembering who that was... but if it was Userbenchmark, then my point stands.

Yup, they were the ones.

Also watch out for AMD’s army of Neanderthal social media accounts on reddit, forums and youtube, they will be singing their own praises as usual.

They are not to be trusted, even if it makes AMD look good. Couldn't whoever has that chip use some relevant benchmark?
 

HurleyBird

Platinum Member
Apr 22, 2003
2,811
1,544
136
Fake benchmark gonna fake.

Was it uploaded that way or did the userbenchmark people go in and change it out of spite? Took quite a long time for anyone to notice if it was the former.

EDIT: Changed since then from "AMD Hype Train" to "AMD Zen4 Hype Train." Userbenchmark somehow sinking to yet another new low with their unprofessionalism.

EDIT2: I first thought you were saying that the results were fake, but I think you were just calling userbenchmark as a whole fake.
 
Last edited:

Thunder 57

Diamond Member
Aug 19, 2007
4,027
6,741
136
Hahaha wow! This has to be intentional. Hey, it's basically free advertising at this point.

Average Bench: 117% (2nd of 1358)
Based on 1 user benchmark. Device: 178BFBFF00A60F11 Model: AMD Eng Sample: 100-000000593-20_Y
Literally within minutes of this sample appearing on userbenchmark, AMD’s marketing machinery was actively promoting Zen 4 whilst simultaneously slandering userbenchmark via hundreds of “news” outlets and thousands of twitter and reddit accounts. If AMD were about to release a class leading CPU (or GPU) they would have no incentive to take this approach. As it stands, buying AMD’s products is similar to buying used cars. It’s difficult for consumers to make rational choices because AMD completely dominate “news” and social media channels. If these marketing practices continue, Ryzen will quickly end up in the same state as Radeon. The combined market share for all of AMD’s Radeon RX 5000 and 6000 GPUs (June’22 Steam stats) is just 2% whilst Nvidia’s RTX 2060 alone accounts for 5%. Of course, if Zen 4 actually delivers anywhere near a 57% single core uplift, we will bow down, call AMD king, and commit seppuku. [Jul '22 CPUPro]
 

Thunder 57

Diamond Member
Aug 19, 2007
4,027
6,741
136
RedGamingTech has acquired some rumored prices. Also, rumored launch target date is Sept. 15th.

7950x = $799
7900x = $549
7800x = $449
7700x = $349
7600x = $299
7600 (non X) = $229


I'm not sure about that. I haven't watched the video but I can't see any reason they would have two 8 core chips priced $100 apart. Are there enough chips available to release a $229 non-x? It will be a whole new platform making cost of entry expensive, so it is possible. Whereas the 5600X was a drop in upgrade for many people and used DDR4 it was a less expensive upgrade for most.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,811
1,544
136
Hahaha wow! This has to be intentional. Hey, it's basically free advertising at this point.

If the 7600X does get a >50% 1T uplift on userbenchmark, it will probably and ironically be because Zen 4 happens to exploit the existing labyrinthine measures userbenchmark has taken to bias their results. Userbenchmark will not "commit seppuku," but will once again predictably change their algorithm to fit a predetermined narrative.
 

Saylick

Diamond Member
Sep 10, 2012
4,036
9,456
136
Was it uploaded that way or did the userbenchmark people go in and change it out of spite? Took quite a long time for anyone to notice if it was the former.

EDIT: Changed since then from "AMD Hype Train" to "AMD Zen4 Hype Train." Userbenchmark somehow sinking to yet another new low with their unprofessionalism.

EDIT2: I first thought you were saying that the results were fake, but I think you were just calling userbenchmark as a whole fake.
Yeah, Userbenchmark apparently changed the weighting to give Intel the lead again... Way to shred any ounce of credibility it had left... Good riddance.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,106
136
Yeah, Userbenchmark apparently changed the weighting to give Intel the lead again... Way to shred any ounce of credibility it had left... Good riddance.
Let's be real; they lost all credibility ages ago. At this point it's just entertainment to see how far off the deep end they can go. Judging by the complete meltdown they're having over this...yeah...
 

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
From WCCFTECH

"Update: UserBenchmark might've been a little butthurt by the fact that a leaked Ryzen chip was scoring better than an upcoming Intel CPU that it decided to rename the entry to "Advanced Marketing Devices"."
 

eek2121

Diamond Member
Aug 2, 2005
3,410
5,049
136
RedGamingTech has acquired some rumored prices. Also, rumored launch target date is Sept. 15th.

7950x = $799
7900x = $549
7800x = $449
7700x = $349
7600x = $299
7600 (non X) = $229


Dude is clueless as usual. Chances are he made a guess based on Zen 3 pricing. Those price points certainly did not come from a “leak”.

Most of those SKUs likely aren’t even launching this year (unless AMD has been secretly building inventory). 5nm capacity will be an issue at first, so likely we will see a launch similar to Zen 3, then later we will get the X3D parts, then next year all the other stuff.
 

moinmoin

Diamond Member
Jun 1, 2017
5,242
8,456
136
Way to shred any ounce of credibility it had left...
The only ounce of credibility Userbenchmark/CPUPro had left before was for believers of its conspiracy theories already. Not sure those believers suddenly see the pure inanity of all of CPUPro's writings.
 
  • Like
Reactions: Saylick