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!

 

[Det0x]

More leaks

AMD Ryzen 7950X/7900X/7700X/7600X stock Cinebench scores leaked - VideoCardz.com

AMD Ryzen 7000 tested in Cinebench Just a day before official review coverage, we were given final scores for Cinebench benchmarks (and much more). This post is a little teaser of what will be shared tomorrow by reviewers from around the world. AMD Ryzen 7000 Desktop CPU Series, Source: AMD...

videocardz.com

In this short article we will share Cinebench R15, R20 and R23 scores for all AMD Ryzen 7000 CPUs in comparison to their 5000 predecessors. All Ryzen 7000 CPUs were tested with stock settings with AMD provided review kit, featuring premium X670E motherboard and G.Skill DDR5-6000 memory (2x16GB). AMD Ryzen 5000 CPUs with DDR4-3200 memory.


*edit*
If other leaks are to be believed, a stock 13900k scores 2290 ST and 35700 MT in Cinebench 23 (250w and unknown cooling)

*edit2*
Do note Cinebench is one of the benchmarks with alittle below average IPC increase for Zen3 vs Zen4

Looking forward seeing the gaming benchmarks tomorrow

https://twitter.com/x/status/1573729834190376960

 

[Asterox]

More leaks

AMD Ryzen 7950X/7900X/7700X/7600X stock Cinebench scores leaked - VideoCardz.com

AMD Ryzen 7000 tested in Cinebench Just a day before official review coverage, we were given final scores for Cinebench benchmarks (and much more). This post is a little teaser of what will be shared tomorrow by reviewers from around the world. AMD Ryzen 7000 Desktop CPU Series, Source: AMD...

videocardz.com

In this short article we will share Cinebench R15, R20 and R23 scores for all AMD Ryzen 7000 CPUs in comparison to their 5000 predecessors. All Ryzen 7000 CPUs were tested with stock settings with AMD provided review kit, featuring premium X670E motherboard and G.Skill DDR5-6000 memory (2x16GB). AMD Ryzen 5000 CPUs with DDR4-3200 memory.
View attachment 68072

*edit*
If other leaks are to be believed, a stock 13900k scores 2290 ST and 35700 MT in Cinebench 23

More clicks, and then everything is more or less pointlessly ordinary Cinebench recycling. You should focus on the tests of everyday normal applications.Cinebench is blah, if you never do any CPU rendering especially not with Maxon products.

Page 33 - Zen 2 APUs/"Renoir" discussion thread

Page 33 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[TESKATLIPOKA]

For some unknown reason, R7 7700x has only 14.5% higher performance in CB R15 MT(2920 vs 2551).

 

[Abwx]

Cinebench is blah, if you never do any CPU rendering especially not with Maxon products.

Page 33 - Zen 2 APUs/"Renoir" discussion thread

Page 33 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

They do, but i dont think that a single user of Cinema 4D would do renderings in ST...

What matter in this bench is the MT score as well as power efficency.

 

[Kaluan]

Oh jeez, not Cinebench again 😂

For some unknown reason, R7 7700x has only 14.5% higher performance in CB R15 MT(2920 vs 2551).

ST as well, it's as if there's no IPC or IPC regression. The 7700X tests in particular are the weirdest.

I wonder what their source is, they didn't even post SSs of anything.

At this point I don't even care about leaks (unless they're PR stunts AKA early publishing "mistakes" like LTT pulled awhile back). Just wait for the good stuff (ComputerBase, TPU, Anand, GN etc) tomorrow.

Anyone have a hour for the NDA lift tomorrow?
Hopefully I have more than one thing to celebrate tomorrow and I can comfim my new build plans.

 

[deasd]

If R23 ~37423 pts is the stock 7950X result, what is the ~40498 pts leaked few days before which showed it only consumed 227 watts?? How much wattage do stock ~37423 pts consume?

 

[Carfax83]

Oh jeez, not Cinebench again 😂

Tell me about it! Has there been any gaming performance leaks yet? I find it odd that there really hasn't been much of any gaming leaks yet.

I saw this on YouTube but I don't know if it's legit, 7600x with DDR5 6200 vs 12900K with DDR4 3600. Don't know why the uploader crippled the 12900K with DDR4 memory though:

 

[Abwx]

If R23 ~37423 pts is the stock 7950X result, what is the ~40498 pts leaked few days before which showed it only consumed 227 watts?? How much wattage do stock ~37423 pts consume?

View attachment 68078

40k+ is an OC with as low as possible voltage, official score will be lower at same TDP because of the necessary voltage margin.

 

[Kaluan]

Tell me about it! Has there been any gaming performance leaks yet? I find it odd that there really hasn't been much of any gaming leaks yet.

I saw this on YouTube but I don't know if it's legit, 7600x with DDR5 6200 vs 12900K with DDR4 3600. Don't know why the uploader crippled the 12900K with DDR4 memory though:

None AFAIK, just some AotS stuff I think, which frankly is worse than CB leak spam.

9 out of 10 times those 'comparison' videos are pure YT clickbait spam from my experience, don't take them seriously.

If R23 ~37423 pts is the stock 7950X result, what is the ~40498 pts leaked few days before which showed it only consumed 227 watts?? How much wattage do stock ~37423 pts consume?

View attachment 68078

Well, I kept saying, just because Ryzen 7000 have x and y TDPs/PPT, that doesn't mean that is EXACTLY (142/229,5W) how much they will use under full load. AVX512 loads may max them out, but there's no reason to believe other tasks will. Case and point 7600X, even tho it's supposed to have a limit of 142W, leaks have already pointed out it only draws between 90 and 120W, depending on the task and microcode/voltage behavior.

Hopefully the review suites come with a 99% solid day 1 BIOSes and microcode. Finewine is nice and all but AMD needs to have a good showing from day 1. As these are the reviews casual builders will come across most. Anyway, this was a bit of a tangent lol

 

[DisEnchantment]

9 out of 10 times those 'comparison' videos are pure YT clickbait spam from my experience, don't take them seriously.

I bet his 'simulated' numbers would be closer to the real deal than MLID's 'sources'.
But either way, if you are into such stuff you can't go wrong with MLID and good ole Mike Benchmark

 

[Henry swagger]

https://uploads.disquscdn.com/images/b6ff33d210a9ac5d84a3201aa86d50b0df7a5d36ae12a57b0384e970676767a4.png

Gaming performance leak.. 58003d and 12900k are still faster

 

[Justinus]

https://uploads.disquscdn.com/images/b6ff33d210a9ac5d84a3201aa86d50b0df7a5d36ae12a57b0384e970676767a4.png

Gaming performance leak.. 58003d and 12900k are still faster

Literally don't believe a 5900X is faster In doom eternal than a 7950X.

 

[Kaluan]

https://uploads.disquscdn.com/images/b6ff33d210a9ac5d84a3201aa86d50b0df7a5d36ae12a57b0384e970676767a4.png

Gaming performance leak.. 58003d and 12900k are still faster

That 'review' looks so full of ** I physically feel constipated just by looking at it.

5700Xs faster than 5950X or tying with 5800X3D, 3900XT being = 5900X, 5600 being much faster than 5600X, general results not lining up with other reputable sites and a plethora of other oddities.

May I reming people, this Spanish language outlet is also the same that broke NDA on Rembrandt (6800H) in the beginning of the year as well. They don't like respecting rules. Wouldn't be so bad if they were at least good a reviews. 😂

 

[moinmoin]

Literally don't believe a 5900X is faster In doom eternal than a 7950X.

The Doom Eternal list looks really odd indeed. Even 7950X aside, can that be real?

this Spanish language outlets

Is its name really "profesional review"?

 

[Kaluan]

Is its name really "profesional review"?

Yes

Profesional Review ⋆ Noticias, Análisis y Guías de Informática

Profesional Review es el blog más importante de noticias de hardware, análisis, gaming, periféricos, smartphone, windows e informática

www.profesionalreview.com

So cheesy (and boast-y) 😅
They're really hard to look up, search engine index really hates them

 

[inf64]

Maybe they left out the "Un" from the website's name . Unprofessional Review sounds about right.

 

[tamz_msc]

Full review from that site (images only):

https://imgur.com/a/d0zcNMd

 

[TESKATLIPOKA]

Full review from that site (images only):

https://imgur.com/a/d0zcNMd

30,000 in CB R23 at 65W(88W PPT)?
Very nice.

 

[tamz_msc]

Yeah, review itself seems legit - only issue I have is the gaming benchmarks. Some of the numbers are way off, as already pointed out by others.

 

[Kaluan]

General inconsistencies aside. If i understand correctly "7950X" in the charts = 7950X w/ JEDEC spec DDR5-4800, while "7950X - DDR5-6000" ...is self-explanatory.

Since I can't read the write-up, I can't confirm the gaming tests were done w/ JEDEC 4800 or otherwise. Either way, some interesting data in there, but overall a confusing mess. I feel like I have even more questions than I had before. 😂

 

[igor_kavinski]

I love that 7950X in ECO mode matches or exceeds the 5950X. Why run it full tilt when you can get excellent everyday performance while also saving on electricity costs and being kind to the environment?

@Kocicak, is this more to your liking?

 

[scineram]

1
getting the most performance per watt out the box

M2 Pro or Max will destroy it in efficiency.

 

[moinmoin]

M2 Pro or Max will destroy it in efficiency.

Actually by Apple's own numbers M2 Pro appears to be about half as efficient as the base M2. But yeah, doubt any Intel chip can beat Apple in efficiency.

 

[Carfax83]

Literally don't believe a 5900X is faster In doom eternal than a 7950X.

I said several pages ago that the Achilles heel of Zen 4 would be it's memory controller, and I based that statement on preliminary memory benchmarks from Aida.

Gaming stresses the CPUs memory controller and interface more than any other consumer oriented workload, so ostensibly, having a weak memory interface would hurt gaming performance.

And it's funny because that Spanish preview used slower DDR5 for the 12900K than the 7950x and the former still outperformed the latter in the memory tests, which indicates that Alder Lake's memory controller is superior to Zen 4's, and also possibly that Alder Lake may be better at hiding latency than Zen 4.

 

[Justinus]

I said several pages ago that the Achilles heel of Zen 4 would be it's memory controller, and I based that statement on preliminary memory benchmarks from Aida.

Gaming stresses the CPUs memory controller and interface more than any other consumer oriented workload, so ostensibly, having a weak memory interface would hurt gaming performance.

And it's funny because that Spanish preview used slower DDR5 for the 12900K than the 7950x and the former still outperformed the latter in the memory tests, which indicates that Alder Lake's memory controller is superior to Zen 4's, and also possibly that Alder Lake may be better at hiding latency than Zen 4.

None of what you said explains why that benchmark had the 5600 faster than the 5900X, 5800X, and 5600X in Doom Eternal.

Again, significant doubt. Those benchmarks don't pass the smell test, period.