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

[Vattila]

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!

 

[Karnak]

Wrong. Except for 7-zip where AMD always had an advantage, the 12600K matches or beats the 5800X in most other multi-threaded apps.

Don't think that is what you call "wrong".

 

[TESKATLIPOKA]

.....
Don't think that is what you call "wrong".

Why didn't you post a bigger picture? I could barely read It. (sarcasm)
Please resize It.
He is partially right, because If you ignored 7-zip, where 5800x is 20% faster, then they would be equal.


edit: added another chart.

 

[inf64]

Why didn't you post a bigger picture? I could barely read It. (sarcasm)
Please resize It.
He is not exactly wrong, because If you ignored 7-zip, where 5800x is 20% faster, then the difference wouldn't be 3% in favor of 5800x.
View attachment 66628

Well if you eliminate the 7zip you still end up with pretty much the same MT performance between the two (+-3% overall doesn't change anything).
Where the difference was more pronounced was ST and this is now addressed with Zen 4.

 

[Ranulf]

5800X's original MSRP was $449. So yeah. $299 would be an admission that Intel's E Core Spam strategy works. As others mention there is no reason to have another 8 core without vcache.

Ignore e-cores for the moment. They don't really matter except as marketing in bigger Core/Thread counts and maybe a few isolated programs. For gaming, they don't matter and probably wont for several more years. Its hyper threading all over again.

$299 is already here because of the 5700X which was brought out for the last 5 months to offer competition to Intel's mid and low end offerings. AMD would be crazy to think given all the factors in play that they could sell another 8 core at $449. Only the most insane fan boys would buy an 8 core cpu at $450-500 after the price drops on cpus this year that were already inflated because AMD had no competition two years ago along with high demand. This is not the case now. It would be an even worse price/perf than the 3000 to 5000 jump was. 50-55% price increase for a 15-25% performance increase? Maybe a 7800X3D can pull off a $450 price point in a few months but thats it.

Toss in expensive motherboards and yeah, thats not gonna fly except for edge cases in a few games.

 

[Vope45]

AMD Ryzen 9 7950X & Ryzen 7 7700X "Alleged" Zen 4 CPU Benchmarks Leak Out, 16 Core Up To 40% Faster Than 5950X

AMD's Ryzen 9 7950X & Ryzen 7 7700X "Zen 4" Desktop CPU benchmarks have been leaked, showcasing over 40% faster multi-thread performance.

wccftech.com

You must post your own commentary along with any link.


esquared
Anandtech Forum Director

 

[DrMrLordX]

If they're priced too high, and Intel undercuts them again, they're going to look really dumb cutting prices 2 months after launch.

I don't think I'm the first to mention it, but Intel is raising prices. AMD knows this, too, and can price accordingly.

 

[PJVol]

For the 7700X these 750-780 in cpu-z just fit in clocks increase. Where is IPC then?

 

[Abwx]

For the 7700X these 750-780 in cpu-z just fit in clocks increase. Where is IPC then?

CPU Z was "updated" when Zen 1 was released under the pretense that it was favouring too much the new AMD uarch, seems that things are working only one way...

 

[PJVol]

CPU Z was "updated" when Zen 1 was released under the pretense that it was favouring too much the new AMD uarch, seems that things are working only one way...

Idk what they "changed" but since Zen2 it scales very good with the max single core frequency.
WTFTech chart show 5800-5950X having the same ST score? What I don't get is how 5950X cannot reach 5050Mhz in single core load? It's ST score should be ~ 705, 5800X should score ~ 675.
Anyway, these scores point to "low single digit") ipc increase, i.e. 0-3% in that particular benchmark.
Something doesn't add up )

 

[inf64]

Idk what they "changed" but since Zen2 it scales very good with the max single core frequency.
WTFTech chart show 5800-5950X having the same ST score? What I don't get is how 5950X cannot reach 5050Mhz in single core load? It's ST score should be ~ 705, 5800X should score ~ 675.
Anyway, these scores point to "low single digit") ipc increase, i.e. 0-3% in that particular benchmark.
Something doesn't add up )

I would just ignore the CPUz for now, especially since we had heard reports that the benchmark was crashing on Ryzen 7000 for whatever reason. The launch and benchmarks are in 3 or so days, we'll have the official numbers very soon. I just hope AMD doesn't mess up the pricing as it could totally ruin this launch. I'm confident the products are great, but out of whack pricing can kill it quickly.

 

[PJVol]

@inf64
Yeah, I understand your concern about pricing and how it will look like and so on, but for me as the particular end user all this doesn't mean a **** for the product which is intended to use 2-3 years, especially on a brand new platform. The only thing that matters to me is its relative performance to the previous gen. Lets hope you're right regarding cpu-z, though.

 

[deasd]

If we look into the Genoa results which Yuuki_Ans posted a month ago, you can see the R20 IPC is significantly better with Raphael,
but the Cpu-z IPC still no better than Milan just like Geona. You can see some results generated by Genoa is much more terrible than Raphael, and experienced crash when doing MT run just like the report by Harukaze5719:

https://twitter.com/x/status/1548717866198831104

https://twitter.com/x/status/1562751758673137664

these dissymmetrical results include Raphael look like a mess to me, even if it's little better.


edit: the result which is marked red is 'in doubt', and 'X' represent crash with unknown reason

 

[inf64]

@inf64
Yeah, I understand your concern about pricing and how it will look like and so on, but for me as the particular end user all this doesn't mean a **** for the product which is intended to use 2-3 years, especially on a brand new platform. The only thing that matters to me is its relative performance to the previous gen. Lets hope you're right regarding cpu-z, though.

For me personally, it will always be a big core only product. I know E cores have their place and can be useful, but I don't like the hybrid approach for my personal use. So it's either full big core product from intel or a full big core product from AMD. Price is important to me and I will not pay 500USD for an 8C/16T product with big cores in 2022, even if it is a stellar performer. I can simply buy 5900X once the price drops more, slot it in my MB easily, and it will be great for gaming and doing work on my PC.

If we look into the Genoa results which Yuuki_Ans posted a month ago, you can see the R20 IPC is significantly better with Raphael,
but the Cpu-z IPC still no better than Milan just like Geona. You can see some results generated by Genoa is much more terrible than Raphael, and experienced crash when doing MT run just like the report by Harukaze5719:

https://twitter.com/x/status/1548717866198831104

these dissymmetrical results include Raphael look like a mess to me, even if it's little better.


edit: the result which is marked red is 'in doubt', and 'X' represent crash with unknown reason

Yeah, I wouldn't pay attention to the "unstable" platform. It's AMD, they are super paranoid with regards to leaks, so all this is "normal". I don't think that any of the ES we had seen in the wild are "full" Zen 4 parts. AMD would never launch a product that cannot finish benchmarks, so all is good

 

[Kaluan]

Wasn't the "source" of that MLID? I'd expect it to be closer to $500.

No, one was WCCFTech and the other was VideoCardz (the one about the new Ryzen 7000 packaging). But I have no doubt MLID also had a go a specu-leaking the prices.
$500? No freaking way. That makes zero sense. Unless I misunderstood and you're talking about the 8-core V-Cache model.

Wrong. Except for 7-zip where AMD always had an advantage, the 12600K matches or beats the 5800X in most other multi-threaded apps.

Intel Core i5-12600K Review: 5600X Defeated

The Core i5-12600K is Intel's latest mainstream CPU and a direct competitor to AMD's popular Ryzen 5 5600X. The Alder Lake chip packs 6 P-cores and 4...

www.techspot.com

It might be the same with 13600K and 7700X.

Well yeah, except nope:

Here's some gaming ones for good measure as well:


Basically what you're trying to say is it's unfair to measure 7zip *decompress* (half the 7zip test) cause it bumps up 5800X averages, but it's totally fine benching both R20 and R23, that bump up 12600K averages. lol
(the above averages feature the above mentioned benches)

Maybe if you add fast DDR5 the average tips consistently in favor for of the i5, but then again, are we still talking about the same value propositions? 5800X + mobo + RAM can already be had for slightly cheaper than a DDR4 12600K combo.

 

[jpiniero]

You're also talking about an significant increase in costs too... core die from N7 to N5 and the IO die from cheap GloFo to N6.

Ignore e-cores for the moment. They don't really matter except as marketing in bigger Core/Thread counts and maybe a few isolated programs. For gaming, they don't matter and probably wont for several more years. Its hyper threading all over again.

I expect the 13700K will be about $500 too, and that's 8 P cores. That's what I mean about the E Core Spam...

Maybe it'll end up being a marketing mistake that they didn't call it the 7800X even assuming there's an 7800X3D, if people are going to get hung up over the 5700X's MSRP.

 

[Asterox]

CPU Z was "updated" when Zen 1 was released under the pretense that it was favouring too much the new AMD uarch, seems that things are working only one way...

It is one more ultimate clickbait CPU benchmark.

 

[Abwx]

Idk what they "changed" but since Zen2 it scales very good with the max single core frequency.
WTFTech chart show 5800-5950X having the same ST score? What I don't get is how 5950X cannot reach 5050Mhz in single core load? It's ST score should be ~ 705, 5800X should score ~ 675.
Anyway, these scores point to "low single digit") ipc increase, i.e. 0-3% in that particular benchmark.
Something doesn't add up )

It scale with CPU brand...

The 5950X should have 7.5% better ST perf than the 5600X and that s far from being the case in the WCCFTech charts.

 

[Hans Gruber]

Wccftech is completely unreliable. I ignore everything they publish. Eg. they said that Zen 4 would be delayed because of bios issues. This was a week ago. Nothing they publish is accurate or truthful.

 

[nicalandia]

Here is my prediction. We can come back to this posts at the end of the year. I think the Raptor Lake CPU wins over Zen 4 in performance by a decent margin in most categories except for power consumption.

Zen3 and Golden Cove have the same gaming performance at ISO Speed and ISO DDR4(see 12400F vs 5600X) due to Zen3 better memory Susbsystems... AMD just reached Speed and RAM Parity at stock so both Z4 and Raptor Cove are enhancements over their previous generations(Zen4 even more due to die shrink, AVX512).

Zen4 overal IPC has gone up(8-10). If Raptor Cove reaches the same IPC uplift this thing is a Neck and Neck race in most production type workloads(AVX512 Will go to Zen4) and Gaming will go to Zen4.

 

[tamz_msc]

Well yeah, except nope:
View attachment 66637View attachment 66638
Here's some gaming ones for good measure as well:
View attachment 66639View attachment 66640

Basically what you're trying to say is it's unfair to measure 7zip *decompress* (half the 7zip test) cause it bumps up 5800X averages, but it's totally fine benching both R20 and R23, that bump up 12600K averages. lol
(the above averages feature the above mentioned benches)

Maybe if you add fast DDR5 the average tips consistently in favor for of the i5, but then again, are we still talking about the same value propositions? 5800X + mobo + RAM can already be had for slightly cheaper than a DDR4 12600K combo.

I'm not saying that it is unfair to measure 7-zip decompress. I'm saying that after a point when you have sufficient performance it doesn't matter because multi-core CPUs are already fast enough to execute it with reasonable speed. People's workflow doesn't involve compressing and decompressing all day.

On the other hand, the only reason we are discussing graphs from Computerbase is because half of their suite consists of render applications, which has been a strong point of AMD since Zen. You can cherry pick all you want, but I don't consider Computerbase suite of benchmarks a representative one because it skews heavily toward one type workload and it doesn't have any benchmark of the Adobe suite at all. People doing work with their computers do not run Cinebench, POV-ray and Corona renderer.

Anyway, to remove bias you should be considering the average of averages across all websites that do benchmarking. Fortunately, 3dcenter has compiled the data for such a comparison, and the result is that the 12600K is slightly faster than the 5800X.

Launch-Analyse Intel Alder Lake (Seite 2) | 3DCenter.org

Sonntag, 7. November 2021 / von Leonidas Im Vorfeld der Alder-Lake-Benchmarks gab es einige Bedenken, ob man so einfach vergleichbare Benchmark-Zahlen würde erreichen können. Schließlich gab es hierzu die Störfaktoren der je nach

www.3dcenter.org

 

[nicalandia]

People's workflow doesn't involve compressing and decompressing all day.

Mine does..

 

[nicalandia]

The 7800X3D will have at least a 20% Speed boost over the 5800X3D... Let that Sink In..!

 

[tamz_msc]

Mine does..

Are you involved with any group that pirates PC games? /jk

 

[Saylick]

The 7800X3D will have at least a 20% Speed boost over the 5800X3D... Let that Sink In..!

Lol, clocks and IPC alone should account for a 20% boost. Don't even need to factor in an improved 3D Cache to make that conclusion.

 

[nicalandia]

Lol, clocks and IPC alone should account for a 20% boost. Don't even need to factor in an improved 3D Cache to make that conclusion.

Speed as in Ghz... 7800X3D will boost to 5.4 Ghz