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

Page 8 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

JustViewing

Member
Aug 17, 2022
159
268
96
AMD can just save costs by using 8P chip + 16E chip in the form of Zen5 + Zen4c. The down bins from Turin and Bergamo sold to consumers to save money. TSMC is no longer offering discounts to even their largest customers.
This may not work as Zen5 likely to have additional instructions added to it. AMD don't want to end up like Intel and have to disable these new instruction for the sake of supporting Zen4c.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,101
136
This may not work as Zen5 likely to have additional instructions added to it. AMD don't want to end up like Intel and have to disable these new instruction for the sake of supporting Zen4c.
I'm not personally convinced it'll happen, but if the differences are small enough, they can probably get away with it. Minor things like RAS shouldn't matter for mainstream desktop.
 

BorisTheBlade82

Senior member
May 1, 2020
667
1,022
136
This may not work as Zen5 likely to have additional instructions added to it. AMD don't want to end up like Intel and have to disable these new instruction for the sake of supporting Zen4c.
While Intel uses entirely different architectures, ZenX and ZenXc are basically the same. At least for Zen4 AMD promised full ISA compatibility. And while the first speculation was that they would combine Zen5 and Zen4c, to me it seems much more likely now that they would combine the same generation after all the bad coverage Intel got.
 
Reactions: Kaluan and Tlh97

Thibsie

Senior member
Apr 25, 2017
806
882
136
While Intel uses entirely different architectures, ZenX and ZenXc are basically the same. At least for Zen4 AMD promised full ISA compatibility. And while the first speculation was that they would combine Zen5 and Zen4c, to me it seems much more likely now that they would combine the same generation after all the bad coverage Intel got.

The rumour from months (years?) ago (but IMO more relevant for laptop market) was that they'd combine a small core and a big core with different ISA BUT...
the different ISA was pretty much intended, an illegal instruction on the small core would wake the big core and the thread would continue execution seamlessly on the big core (but maybe back to the small one depending situation). Seems this also imlied shared register (yeah, I know).

This meant that one can't say thee're two cores. Not really. This rather is a hybrid of two cores but only one is ever active at a time.
That was the rumour circulating at the time.
How much of this was really intended as a real product, who knows ?
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,101
136
While Intel uses entirely different architectures, ZenX and ZenXc are basically the same. At least for Zen4 AMD promised full ISA compatibility. And while the first speculation was that they would combine Zen5 and Zen4c, to me it seems much more likely now that they would combine the same generation after all the bad coverage Intel got.
I really doubt either AMD or Intel care about some of the internet whining about hybrid, especially when half of it seems to only be because it's Intel doing it. The actual results speak for themselves.

The main barrier to combining e.g. Zen 5 + 5c is that AMD would have to have the silicon for both ready at the same time. It might be possible, but it would be vastly easier from a development standpoint to stagger the two.
The rumour from months (years?) ago (but IMO more relevant for laptop market) was that they'd combine a small core and a big core with different ISA BUT...
the different ISA was pretty much intended, an illegal instruction on the small core would wake the big core
I really, really doubt that's the direction AMD goes. Not going to say it's outright impossible, but it seems like a massive overhead. Would be a nightmare from the OS's perspective. That "rumor" reads more like uninformed speculation.
 

Thibsie

Senior member
Apr 25, 2017
806
882
136
I really, really doubt that's the direction AMD goes. Not going to say it's outright impossible, but it seems like a massive overhead. Would be a nightmare from the OS's perspective. That "rumor" reads more like uninformed speculation.

Agreed except for the OS point. The thing is the OS is not NOT seeing two cores. The OS is NOT aware of all this. This was supposed to be dealt with IN the CPU intself.

How realistic, dunno. But would eliminate the OS as a variable.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,101
136
Agreed except for the OS point. The thing is the OS is not NOT seeing two cores. The OS is NOT aware of all this. This was supposed to be dealt with IN the CPU intself.

How realistic, dunno. But would eliminate the OS as a variable.
I mean, sure, if you're willing to literally burn an entire extra CPU core, maybe you can simplify things somewhat. But that's a frankly absurd tradeoff for anyone to suggest, much less for merely ISA compatibility.

Very early hybrid implementations in mobile were vaguely similar, with only one cluster active at a time. We've moved away from that for a reason.
 

moinmoin

Diamond Member
Jun 1, 2017
4,993
7,763
136
The rumour from months (years?) ago (but IMO more relevant for laptop market) was that they'd combine a small core and a big core with different ISA BUT...
the different ISA was pretty much intended, an illegal instruction on the small core would wake the big core and the thread would continue execution seamlessly on the big core (but maybe back to the small one depending situation). Seems this also imlied shared register (yeah, I know).

This meant that one can't say thee're two cores. Not really. This rather is a hybrid of two cores but only one is ever active at a time.
That was the rumour circulating at the time.
How much of this was really intended as a real product, who knows ?
That's not a rumor but an actual patent AMD filed quite some time ago (edit: filled on 10/27/2017, published on 06/30/2020)
and we previously discussed on this board:
It's also a promising(-ly simple) approach to avoiding having to have cores all be full fat in a hybrid configuration as well as accelerating simple instructions at a very early stage even before being processed for specific cores etc.
 

eek2121

Diamond Member
Aug 2, 2005
3,042
4,259
136
Zen+ has actually 5% better IPC than Zen.

If we exclude games and using Zen 1 as basis the evolution is 21%, 34% and 51% respectively for Zen 2/3/4, Zen 4 brought a slightly better IPC improvement than Zen 3 vs Zen 2.

Btw Zen 4 has 2% better MT IPC than ADL/RPL on a 8C comparison basis.


Zen+ was always an interesting release. AMD claimed IPC improvements, but these were improvements Threadripper 1xxx already had. Where did they come from? The clocks came from slightly improved process.
 

Abwx

Lifer
Apr 2, 2011
11,162
3,858
136
Zen+ was always an interesting release. AMD claimed IPC improvements, but these were improvements Threadripper 1xxx already had. Where did they come from? The clocks came from slightly improved process.

Zen 1 was somewhat a rushed design as they were in dire need to get out of financial losses, likely that to have a rock stable chip they gave up the equivalent of 5% clock cycles to the various latencies of the whole pipeline, so this improvement was already there in the first silicon, just not reliably implementable.
 

Kaluan

Senior member
Jan 4, 2022
503
1,074
106
Based on Angstronomics' claim that Strix Point will be announced in January 2024/CES and MLID's claim that Turin is H1 2024 (which I only take seriously because it lines up with the former), what are the odds of a Q1-Q2 2024 Granite Ridge/Ryzen 8000 launch? AMD does seem to be pushing more and more stuff in shorter timeframes. Although I'll admit the Ryzen 5000 near-2 year old freeze is the outlier, not the new normal for AMD, we'll probably be going back to the Zen-Zen3 cadence or faster. Especially if Intel manages to finally pull another Tiger Lake/"10nm SF" with Meteor Lake/"Intel 4" (still doubtful for now tho).

This may not work as Zen5 likely to have additional instructions added to it. AMD don't want to end up like Intel and have to disable these new instruction for the sake of supporting Zen4c.
Was about to say the same thing.
 

yuri69

Senior member
Jul 16, 2013
434
716
136
Based on Angstronomics' claim that Strix Point will be announced in January 2024/CES and MLID's claim that Turin is H1 2024 (which I only take seriously because it lines up with the former), what are the odds of a Q1-Q2 2024 Granite Ridge/Ryzen 8000 launch? AMD does seem to be pushing more and more stuff in shorter timeframes. Although I'll admit the Ryzen 5000 near-2 year old freeze is the outlier, not the new normal for AMD, we'll probably be going back to the Zen-Zen3 cadence or faster. Especially if Intel manages to finally pull another Tiger Lake/"10nm SF" with Meteor Lake/"Intel 4" (still doubtful for now tho).
AMD "releases" a new APU gen each CES. CES 2023 is Zen 4-based Phoenix. We don't have info about any "Zen 4+"-based APU. Thus the 2024 APU should be the Strix Point.

We are nearing end of 2022 but we still have very little info about Zen 5. Is the CPU Family known, yet? My guess is 20h, but was it confirmed? The same goes for the rest of the Zen 5 product line - do we have a model number?

If so, this would imply the initial NDAed docs for Zen 5 products are not yet available outside AMD. This includes server products like Turin, which is kinda strange for "early 2024".

FYI Genoa family + model number got leaked in Sep 2020. This means more than two years prior launch.
 

Kaluan

Senior member
Jan 4, 2022
503
1,074
106
Here is old rumor but a news to here that Zen5 is not big-little. IIRC he is Greymon on twitter. He got info from AMD package facility.



View attachment 70171


I have a wild speculation that 2 CCDs have been unified as one on die package level.
Does just changing cache sizes count among cores on the same package as "big.LITTLE" core setups? Pertinent question IMO. Pretty sure AMD isn't interested in "LITTLE" cores in the ARM/Intel sense anyway.

Interesting bit about the Navi3x, I guess the 2GCD+8MCD future SKU rumor is still alive and well.
 

Geddagod

Golden Member
Dec 28, 2021
1,201
1,164
106
Does just changing cache sizes count among cores on the same package as "big.LITTLE" core setups?
I would say so. You still have different performance/area levels and have to deal with all the software scheduling that goes with such designs.
Pretty sure AMD isn't interested in "LITTLE" cores in the ARM/Intel sense anyway.
I don't think Intel and ARM are doing little cores in the same way. ARM is for improving efficiency, esp when big cores aren't needed, but Intel is using little cores to improve performance, afaik. Which is why the number of little cores for processors such as Apple chips aren't really increasing, but Intel doubled e-cores for raptor lake. I think AMD is going to use little cores more akin to the Intel method.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,101
136
Pretty sure AMD isn't interested in "LITTLE" cores in the ARM/Intel sense anyway.
I think AMD's definitely trending more towards and Intel-like solution than an Apple one. Not saying they will, but if AMD actually does use a last-gen/cut-down core paired with the current, full fat one, then that would probably be a similar performance gap compared to Atom vs Core.
 

Doug S

Platinum Member
Feb 8, 2020
2,470
4,028
136
I would say so. You still have different performance/area levels and have to deal with all the software scheduling that goes with such designs.

I don't think Intel and ARM are doing little cores in the same way. ARM is for improving efficiency, esp when big cores aren't needed, but Intel is using little cores to improve performance, afaik. Which is why the number of little cores for processors such as Apple chips aren't really increasing, but Intel doubled e-cores for raptor lake. I think AMD is going to use little cores more akin to the Intel method.


Apple & ARM little cores are primarily about saving power. Intel's little cores are primarily about saving area.
 

Joe NYC

Platinum Member
Jun 26, 2021
2,323
2,929
106

moinmoin

Diamond Member
Jun 1, 2017
4,993
7,763
136
Apple & ARM little cores are primarily about saving power. Intel's little cores are primarily about saving area.
And AMD's mobile cores so far have been about saving power, saving area and higher density, but not changing ISA features.

The low-end budget Mendocino will actually be the first publicly sold Zen chip that cut capability beyond just different L3$ sizes, with only the semi-custom Zen 2 cores in PS5 consoles predating that.
 
Reactions: Kaluan
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |