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

Page 334 - 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
805
1,390
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).



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:
Reactions: richardllewis_01

coercitiv

Diamond Member
Jan 24, 2014
6,343
12,602
136
I don't think Zen 4 V-cache will be giving you +20% in gaming across a large swathe of games. Yes, there will be some instances where the gains are very large, but overall DDR5 gains should somewhat negate the effect of additional V-cache.
The discussion we had earlier in this thread pivoted around Zen 3 V-cache showing 15% gains in gaming while running with a 9% clock deficit. Even if we apply a 2/3 diminishing return ratio from DDR5, eliminating the clock deficit through better stock behavior + overclocking would still go past 15%+ gains. Keep in mind that most of the gains from DDR5 should be absorbed by the faster vanilla Zen 4 performance, we're not just talking abut Zen 3 with shiny new DDR5 capabilities.

For the sake of the argument though, let's consider a more conservative 10-15% gain from V-cache with no clock deficit. That's still 20-25% faster than 12900K /w DDR5 6000. Quite a mouthful for Raptor OC.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,859
3,726
136
For the sake of the argument though, let's consider a more conservative 10-15% gain from V-cache with no clock deficit. That's still 20-25% faster than 12900K /w DDR5 6000. Quite a mouthful for Raptor OC.
Well the 13900K is supposed to have 11% higher fmax, 1.6x bigger L2 and much improved L3. From cache improvements alone I expect ~8% higher performance, so assuming everything scales linearly, the 13900K should also be around 20% faster than the 12900K.
 

exquisitechar

Senior member
Apr 18, 2017
664
883
136
The discussion we had earlier in this thread pivoted around Zen 3 V-cache showing 15% gains in gaming while running with a 9% clock deficit. Even if we apply a 2/3 diminishing return ratio from DDR5, eliminating the clock deficit through better stock behavior + overclocking would still go past 15%+ gains. Keep in mind that most of the gains from DDR5 should be absorbed by the faster vanilla Zen 4 performance, we're not just talking abut Zen 3 with shiny new DDR5 capabilities.

For the sake of the argument though, let's consider a more conservative 10-15% gain from V-cache with no clock deficit. That's still 20-25% faster than 12900K /w DDR5 6000. Quite a mouthful for Raptor OC.
Well the 13900K is supposed to have 11% higher fmax, 1.6x bigger L2 and much improved L3. From cache improvements alone I expect ~8% higher performance, so assuming everything scales linearly, the 13900K should also be around 20% faster than the 12900K.
Whichever one of them ends up in the lead, I think the difference will be marginal on average. I think the regular Raphael will be firmly behind both, however. AMD probably doesn’t care about that since the difference will be small in most popular reviews.
 
Reactions: Zucker2k

TESKATLIPOKA

Platinum Member
May 1, 2020
2,409
2,904
136
For efficiency, they compared 7950x vs 5950x in CB R23 nT.

Footnotes for both slides:

This slide is straightforward. I am happy with much higher performance at low TDP, that means mobile will see significant improvement in performance.


This slide doesn't tell us much info to compare.
In the case of Up to ~49% higher performance(same power), we don't know what TDP they are actually talking about. It must be somewhere between 65-105W TDP based on the increase in performance and looking at the first slide.
In the case of Up to ~62% lower power(same performance), we don't know what is the R23 score they are talking about.
I personally think they are talking about R23 score of 5950x at 170W TDP(230W PPT).
This would also mean 7950x needs only 65W TDP(88W PPT) for the same performance. Why do I think so? Because Zen4 is most efficient at 65W TDP(88W PPT) based on the first slide.

BTW I checked 5950x review(Link) at ComputerBase. They didn't test CB R23, but at least they tested CB R20, there shouldn't be much of a difference.
If I apply the first slide to CB R20 scores, then It looks like this for 7950x.
CB R20TDP: 65W
(PPT: 88W)
TDP: 105W
(PPT: 142W)
Performance gain from
higher TDP
5950x79821043531%
7950x13889 (+74%)14296 (+37%)3%
This doesn't look correct, only 3% difference in performance with 62% higher TDP(65->105W)? Nonsense.
 
Last edited:

Vope45

Member
Oct 4, 2020
114
168
86
Well the 13900K is supposed to have 11% higher fmax, 1.6x bigger L2 and much improved L3. From cache improvements alone I expect ~8% higher performance, so assuming everything scales linearly, the 13900K should also be around 20% faster than the 12900K.
Sorry to disappoint you but raptor lake does not matter at all. For alder lake users, raptor is a very very small side grade. For all other users, amd offer a much much better core and platform. Hope this helps.
 

exquisitechar

Senior member
Apr 18, 2017
664
883
136
For efficiency, they compared 7950x vs 5950x in CB R23 nT.

Footnotes for both slides:
View attachment 66795
This slide is straightforward. I am happy with much higher performance at low TDP, that means mobile will see significant improvement in performance.


This slide doesn't tell us much info to compare.
In the case of Up to ~49% higher performance(same power), we don't know what TDP they are actually talking about. It must be somewhere between 65-105W TDP based on the increase in performance and looking at the first slide.
In the case of Up to ~62% lower power(same performance), we don't know what is the R23 score they are talking about.
I personally think they are talking about R23 score of 5950x at 170W TDP(230W PPT).
This would also mean 7950x needs only 65W TDP(88W PPT) for the same performance. Why do I think so? Because Zen4 is most efficient at 65W TDP(88W PPT) based on the first slide.

BTW I checked 5950x review(Link) at ComputerBase. They didn't test CB R23, but at least they tested CB R20, there shouldn't be much of a difference.
If I apply the first slide to CB R20 scores, then It looks like this for 7950x.
CB R20TDP: 65W
(PPT: 88W)
TDP: 105W
(PPT: 142W)
Performance gain from
higher TDP
5950x79821043531%
7950x13889142963%
This doesn't look correct.
I can’t wait to see Zen 4 in laptops, it seems like it will be amazing. The RDNA3 iGPU won’t hurt, either.
 

Kuiva maa

Member
May 1, 2014
181
232
116
A lot of bugs guys, a lot of bugs. Not as bad as zen 1, but it's bad.
Zen 1/X370 was a unique situation. AMD debuted a brand new architecture using a brand new chipset, in what was essentially its first real, large volume attempt with AM4, for the first time using DDR4. It was a perfect storm for bugs and yet I was very lucky back then. With Crosshair VI hero being compatible with b die 3200/CL14 from day 1, I was outperforming all the review samples with my also day 1 1800X. Especially in games. So teething issues don’t scare me, it’s the unknown factor of if/when X3D models will arrive and what will they be. I have a 5900X and target yet again the 12C range.
 
Reactions: Thunder 57 and ftt

tamz_msc

Diamond Member
Jan 5, 2017
3,859
3,726
136
Sorry to disappoint you but raptor lake does not matter at all. For alder lake users, raptor is a very very small side grade. For all other users, amd offer a much much better core and platform. Hope this helps.
Alder Lake users aren't going to upgrade to Raptor Lake, true, but don't just assume that the only ones interested in Raptor Lake are those running Alder Lake. On the Intel side, most people do not upgrade the CPU alone - platform upgrades are the norm for the majority.

13th gen will be the undisputed king in the sub-$300 market though. For perf/$, Intel has AMD beat in the mainstream market, because of lower platform cost, DDR4 and E-core spam as some like to call it.
 
Reactions: krumme

AtenRa

Lifer
Feb 2, 2009
14,002
3,357
136
5,7GHz is not Fmax, it's the advertised boost clock. Current rumor is Fmax on that SKU is 5,85GHz.

You're kidding, right? 5600 and 5600X are very competitive agaist what Intel has bellow 12600K. Up to $300? You know you can get 5700X and 5800X for $260, right?
As for 13th gen, the entire lower tier lineup (which won't come up until next year...) is supposedly based on Alder Lake silicon, so don't expect as much as you would from i5 Ks and up.

That being said, the supposed pricing gives me mixed feelings. I think 7600X should be $250-280, 7700X $300-350 and 7900X 500. 7950X price seems pretty good however. 13900K will cost around the same.


1. No competition at $100 vs i3 12100
2. 12600K currently at $239 that gives higher ST and MT vs the 5600X at $200
3. At the $300 range i7 12700F is currently at $320 that decimates the $260 5700X in both gaming and MT workloads.

Plus socket 1700 platform has higher upgradability (Raptor Lake), AM4 is dead socket.

So in October the new Ryzen 7600X at $299 will have to compete against AL 6+4 at $200 (13400). 13400 system will be even cheaper because the platform also supports DDR4.

So it comes down to this, 7600X at $299 will have to compete against i7 13xxx and not i5s when adding the cost of the platform and DDR5 memory.
And again at the sub $200 range AMD will have no answer against Intel, unless they will cut prices and make 5700X compete at the $200-220 segment. Which again will have hard time compete in gaming against the $200-220 Intel CPUs.

And good luck finding a cheap B650 motherboard this year.
 
Reactions: Ranulf and krumme

Vope45

Member
Oct 4, 2020
114
168
86
Alder Lake users aren't going to upgrade to Raptor Lake, true, but don't just assume that the only ones interested in Raptor Lake are those running Alder Lake. On the Intel side, most people do not upgrade the CPU alone - platform upgrades are the norm for the majority.

13th gen will be the undisputed king in the sub-$300 market though. For perf/$, Intel has AMD beat in the mainstream market, because of lower platform cost, DDR4 and E-core spam as some like to call it.

I don't know about that to be honest. Main stream market is going away in favor of performance. Consumers are willing to pay for markup. Look at ampere vs rdna2.
 

AtenRa

Lifer
Feb 2, 2009
14,002
3,357
136
Sorry to disappoint you but raptor lake does not matter at all. For alder lake users, raptor is a very very small side grade. For all other users, amd offer a much much better core and platform. Hope this helps.

There will be many current AL users that Raptor Lake will be a significant upgrade path. For those currently using 12100/12400 they could upgrade to i7 RL and get tremendous increase in performance both in gaming and MT workloads just by upgrading their existing platform with a new CPU swap.

For a new system build, it may be better to go with ZEN4, but I will wait until the end of September to see both ZEN4 and RL and then make any final conclusions.
 
Reactions: krumme

AtenRa

Lifer
Feb 2, 2009
14,002
3,357
136
I believe that AMD can come up with interesting cheap monolithic designs on TSMC 5 nm process both for AM4 and AM5. They could possibly even use exactly the same dies intended for laptops just clocked higher.

There will be no more new AM4 CPUs from now on, especially at 5nm.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,725
1,342
136
That 13% IPC figure truly screams for a SPEC validation. Seeing benches with <5% improvement leads one to doubt. During the Zen 3 showcase the min gain was 9%. For Zen 4 it's 1%.

It wouldn't surprise me at all if something like the 13% average uplift was verified given historical precedent, but still, the technical marketing drone in charge of picking which applications to sample for IPC calculations should really rethink his approach. Besides the gaping lack of a code compilation bench, and the inclusion of some benchmarks of questionable utility (eg. CPU-Z, Deus Ex:MD), this should really be two slides: one for games alone, and the other for general applications.
 

poke01

Golden Member
Mar 8, 2022
1,213
1,398
106
AMD used Geekbench to compare with Intel regarding single core. Now, I don't want to see people saying Geekbench favours Intel and not AMD.

It's not that Geekbench favours Intel but AMD was really behind in ST compared to 12th gen Intel and GB showed that accurately. Zen 4 ups ST for AMD.

So I agree with AMD and say that Geekbench is a great benchmark for singlecore testing for cross platform CPU.
 

Kuiva maa

Member
May 1, 2014
181
232
116
1. No competition at $100 vs i3 12100
2. 12600K currently at $239 that gives higher ST and MT vs the 5600X at $200
3. At the $300 range i7 12700F is currently at $320 that decimates the $260 5700X in both gaming and MT workloads.

Plus socket 1700 platform has higher upgradability (Raptor Lake), AM4 is dead socket.

So in October the new Ryzen 7600X at $299 will have to compete against AL 6+4 at $200 (13400). 13400 system will be even cheaper because the platform also supports DDR4.

So it comes down to this, 7600X at $299 will have to compete against i7 13xxx and not i5s when adding the cost of the platform and DDR5 memory.
And again at the sub $200 range AMD will have no answer against Intel, unless they will cut prices and make 5700X compete at the $200-220 segment. Which again will have hard time compete in gaming against the $200-220 Intel CPUs.

And good luck finding a cheap B650 motherboard this year.

While I fully agree than AMD needs a better value proposal in the very low end I have to point out that comparing prices to make global conclusions is tricky. Take Finland for example. A 12600K can be had for about 310, a 12700F at 410 a 5600X at about 230 and 5800X3D at 500. So it seems to me that AMD is very well positioned here (a decent budget option and a heavy hitter only 90€ above the completion)and the 7600X will add another choice in there. As for 13 series we will have to see first how it stacks up with Zen4 before we make confusions.
 

Vope45

Member
Oct 4, 2020
114
168
86
There will be many current AL users that Raptor Lake will be a significant upgrade path. For those currently using 12100/12400 they could upgrade to i7 RL and get tremendous increase in performance both in gaming and MT workloads just by upgrading their existing platform with a new CPU swap.

For a new system build, it may be better to go with ZEN4, but I will wait until the end of September to see both ZEN4 and RL and then make any final conclusions.

I think 12100/12400 users need to buy new mobo for the new i7s

AMD used Geekbench to compare with Intel regarding single core. Now, I don't want to see people saying Geekbench favours Intel and not AMD.

It's not that Geekbench favours Intel but AMD was really behind in ST compared to 12th gen Intel and GB showed that accurately. Zen 4 ups ST for AMD.

So I agree with AMD and say that Geekbench is a great benchmark for singlecore testing for cross platform CPU.

False. Amd use whatever benchmark that suits them the most. At the moment, there isn't a single benchmark that can capture the whole performance spectrum of zen. I think, to amd, bg is the most appropriate because of avx512.
 

poke01

Golden Member
Mar 8, 2022
1,213
1,398
106
While I fully agree than AMD needs a better value proposal in the very low end I have to point out that comparing prices to make global conclusions is tricky. Take Finland for example. A 12600K can be had for about 310, a 12700F at 410 a 5600X at about 230 and 5800X3D at 500. So it seems to me that AMD is very well positioned here (a decent budget option and a heavy hitter only 90€ above the completion)and the 7600X will add another choice in there. As for 13 series we will have to see first how it stacks up with Zen4 before we make confusions.
yikes everything is expensive in europe. In Australia the 12700F is 360 euros including tax/GST.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
I was expecting Zen 4 be much more modest, something between Zen+ and Zen 2, good enough but a bit disappointing. So for me this whole reveal felt like "big if true".

One more month until we discover if it's all true.

And stop crying about the prices!
The prices show there is now very very little compettition. Like back when Intel sold 4c forever. At least consumers now get solid improvements. But they are still getting profitized from, as companies do.

I have been on this forum when Intel was all the rage, and Intel prices were defended. The forum was worse back then partly due to that. Intel enthusiast defending the price and glorifying Intel. Dont be like that now.
A base level price for a cpu at 300usd is just plain high. Full stop. And it can be seen at the valuation of amd. This industry is haunted by the tendency towards natural monopoly and the resulting high prices.
As enthusiast we should fight that. Not support it.
We have learned that by history.
 
Last edited:
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/    |