• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 6000)

Page 43 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

What do you expect with Zen 4?


  • Total voters
    242

DisEnchantment

Senior member
Mar 3, 2017
811
1,995
136
I can't help but wonder if there is some truth to this report from RGT. Although I don't really like this kind of reporting, Paul has been right many times so I am willing to consider this could be a possibility.
There are no events listed in AMD's IR calendar, total silence til mid of the year. So we wont be finding out any time soon.
I agree on the reasoning as well, Zen 4 on 5nm would use the new Fab 18 which does not share anything with other fabs using N6/N7. N6 and N7 share equipment so I guess capcaity remains an issue if they use N6.
 

Mopetar

Diamond Member
Jan 31, 2011
5,785
2,553
136
Paul reports on a lot of rumors so he's right a lot only because he reports on the rumors that pan out as much as the ones that don't.

If Zen 4 is well along and AMD suspects that Intel is going to have a good showing with their future products there's little reason to devote time to a stop-gap product.

The other side of this could be that AMD was expecting more from Intel in their most recent release and had prepared Zen 3+ as a potential response to regain supremacy. That no longer seems necessary as AMD's biggest issue is being able to supply demand and Zen 3+ does nothing to help that.
 

scineram

Member
Nov 1, 2020
136
100
76
Indeed the reasoning does make sense. But only if it’s for AM4 still, not launching AM5. Which would be the disappointment scenario for me anyway.
 

Kedas

Senior member
Dec 6, 2018
267
223
86
If Alder Lake would only be available on their new DDR5 platform then there is no point for AMD to release an Zen3+ on DDR4 to compete with it.
AMD would still have the best DDR4 platform.

And then we don't even talk about the fact that AMD is selling everything they can produce. So what would change after Zen3+ release, nothing.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,297
723
136
If Alder Lake would only be available on their new DDR5 platform then there is no point for AMD to release an Zen3+ on DDR4 to compete with it.
AMD would still have the best DDR4 platform.
If Zen3+ beats Alder Lake, but Zen3 doesn't (or if it moves the needle from questionable to not), then I disagree.

There's something significant about a last-gen platform beating the next-gen platform of your competitor. And AM4/DDR4 will be around in the market for quite a while. Making the final product line on that platform perform as well as possible would hardly be irrelevant.

Which isn't to say that focusing on Zen4 isn't the right move even if there are reasons for Zen3+. If AMD can speed up Zen 4 development by say, three months, then that sounds like a good trade-off. There's nothing stopping AMD from restarting work on Zen3+ after Zen 4 launches either, if it seems like a good market move and there's already a good chunk of sunk cost.
 

mikk

Diamond Member
May 15, 2012
3,134
946
136
I'm expecting no more than a slighty higher clocked Zen 3 refresh. Maybe they have realized that using a new generation naming for a slightly higher clocking Zen 3 is not a good idea.
 
  • Like
Reactions: Mopetar

Ajay

Diamond Member
Jan 8, 2001
8,556
3,342
136

sounds interesting....

"In addition, the new CPU in the second half of this year will be very interesting."

"Yes, very powerful. Especially in games."

would could that be ??
Uh, this guy has been sure of Warhol as early as April 10th:

Warhol makes sense to me, unless Zen4 is due out sooner than we expect. An important point is that this decision would have been made quite some time ago. AMD isn't making decisions on whether or not to design and make masks for a given chip on how well a particular Intel CPU is performing. They are making decisions based on their own internal time frames; perhaps with some regard to Intel's roadmap (which they know better than we do).
 
  • Like
Reactions: Tlh97 and Elfear

DisEnchantment

Senior member
Mar 3, 2017
811
1,995
136
IMO not a big.little as we know it.
Patents show (if those are indeed used) a quite smart way of implementing this.

Yeah I did write about this patent somewhere back

A more consequential patent is this actually, which I also posted some pages back

US Patent has been awarded was filed on Oct 27 2017

But what is interesting is that AMD applied for another similar patent again. It seems likely that they found additional things to do the with it and applied for it again. Filed June 25 2020.
Not yet awarded, still in application state but will surely be awarded as it is just a continuation of awarded patent.

Similar abstract but 20 additional claims.

You can read both, they are awesome.

View attachment 37823View attachment 37824



Like I wrote in my earlier post, there are shared registers between the big and small cores.
In low power mode the small cores are running only a small subset of low powered instructions, when a complex instruction is encountered a trap occurs and the big core takes over seamlessly.
OS is not even aware that all of this is happening. It is basically the same core to the OS.
Because of shared Registers L2 etc, the number of small cores is exactly the same as the number of big cores and at any time either the big or the small core is active not both.
In my opinion this could be done fairly cheaply in terms of die real estate, they could power gate selected blocks in the current core and add special execution ports, L/S ports and other tidbits and thats all.
I imagine in low power the first target would be to power gate the complex decoder block, leaving just simple decoding blocks, power gate all the executions ports except the most simple and power efficient one, power gate a chunk of the register file, and so on.
More innovative than the big.LITTLE imo.

I suppose this opens up new possibilities on what the CPU can do.
It can execute multiple ISA with the same core :blush: . illegal opcode trap in one core changes the CPU to another and continue execution. And OS is not aware.
Dreams...
I do find their solution to be unique and if there are several follow up patents it usually end up happening.
 

Ajay

Diamond Member
Jan 8, 2001
8,556
3,342
136

Bullsh1t_Buster also says Warhol has been cancelled.
+ 100-200 MHz will be kind of worthless 5000 XTs. So, it's really just a marketing exercise at this point if true. Also, this means Warhol was cancelled something like 2 years ago.

Also, boo hoo, was hoping for a good AM4 replacement for my 3900X. Guess I'll be waiting 2-3 years since I'll need a new board and ram as well as a new CPU.
 

moinmoin

Platinum Member
Jun 1, 2017
2,493
3,144
136
We simply know nothing. Chances are all the CPU rumors, if founded on something real anyway, are from the time when AMD was still more leaky, so well before the launch of the whole Zen 3 line. The last actual data we got from leaks were the specs for RDNA2 chips due to driver and firmware data from several sources. Even that has dried up since.
 

CakeMonster

Golden Member
Nov 22, 2012
1,028
96
91
Which isn't to say that focusing on Zen4 isn't the right move even if there are reasons for Zen3+. If AMD can speed up Zen 4 development by say, three months, then that sounds like a good trade-off. There's nothing stopping AMD from restarting work on Zen3+ after Zen 4 launches either, if it seems like a good market move and there's already a good chunk of sunk cost.
I would consider it very unlikely that it can be sped up, I'd rather assume that they push the release back to build up inventory. I'd love to be wrong though.

I guess we've been conditioned to not expect yearly releases, which is why I never really believed in "Zen3+". However, if Zen4 is late '22 or pushed into '23 then I would expect something IF Intel is also late. We don't really have the info to assess the DDR5 or Alder Lake situation so we're really in the dark here. I feel less sure about making guesses than in a long time. I guess that's why people are hanging on to people posting fanfic on Twitter and YT too.
 

jpiniero

Diamond Member
Oct 1, 2010
9,310
1,869
136
If AM5 only supports DDR5, yes that could be a problem if it is slipping and/or more expensive than even anticipated.
 

uzzi38

Golden Member
Oct 16, 2019
1,589
3,096
106

Bullsh1t_Buster also says Warhol has been cancelled. Interestingly he says AMD is working on a big+little generation. I think it's the future, Intel is just the first with Alder Lake.
This article is written by the owner of that Twitter account: https://moepc.net/amd-zen-5-codenamed-strix-point-will-take-the-big-little-architecture-using-the-3nm-process/

Few extra details that weren't in the Tweets.
 

Ajay

Diamond Member
Jan 8, 2001
8,556
3,342
136
  • Haha
Reactions: scineram

eek2121

Senior member
Aug 2, 2005
965
1,052
136
+ 100-200 MHz will be kind of worthless 5000 XTs. So, it's really just a marketing exercise at this point if true. Also, this means Warhol was cancelled something like 2 years ago.

Also, boo hoo, was hoping for a good AM4 replacement for my 3900X. Guess I'll be waiting 2-3 years since I'll need a new board and ram as well as a new CPU.
For what it is worth, AMD appears to have Warhol as a stand-in for possible 5nm capacity issues OR delayed DDR5 availability. DDR5 looks to be “on track”. Warhol was said to be a DDR4 product and contrary to popular belief, the new die and the old die would not easily interchangeable.

The cancelation rumors may also be completely wrong.

For what it is worth, Only 2 folks have said that it is canceled thus far, and at least one of them is a clown.l and could have gotten it from the other or vice versa.
 

ASK THE COMMUNITY