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!

 

[nicalandia]

Videocardz: AMD Ryzen 9 7950X overclocked to 6.7 GHz scores 50K in Cinebench R23

Scores 48235 in CB R23 at 6.45GHz with power consumption of only 271W.
Power consumption is very low for that clockspeed.

or

Scores 50395 in CB R23 at 6.7GHz with unknown power consumption.

Sweet Jesus...

 

[naad]

I seriously doubt those power figures at those clocks, more than likely sensors not reading it properly, wish we had more power testing directly from the rails

 

[Kaluan]

I seriously doubt those power figures at those clocks, more than likely sensors not reading it properly, wish we had more power testing directly from the rails

Sure but you also have to remember thermodynamics behave differently under sub-zero environmens. That 40K LN2 12900KS at ~7GHz would probably draw much more than 400W under a hypothetical conventional cooling system as well.

Either way, exotic cooling/record run power consumption is irrelevant outside of scientific discussions.

 

[Mopetar]

Lol, only 270W with 6.45ghz all core for 48000 score in CB R23.

Hm if legit, that is far crazy than Klingon overclocking CPU.

i9 13900K needs 350W for 40000 in CB R23.

I seriously doubt those power figures at those clocks, more than likely sensors not reading it properly, wish we had more power testing directly from the rails

More such attempts will bring more data and more insight into what's really going on, but I'm assuming a golden sample combined with some extreme cooling.

Generally as the temperature increases more voltage is needed for stability so keeping a chip running cooler can help a lot. We know that Zen 4 is pretty dense and there's already been a lot of talk about hot spots and cooling challenges.

Liquid nitrogen is probably sufficient to keep those hot spots cool allowing the chip to be pushed a lot harder than it can under air.

 

[nicalandia]

Come on Guys it's just using 1.4 Volts How on earth is it going to draw more than 280 Watts at those voltages?

 

[nicalandia]

Sweet Jesus...

Just saw that Tech Jesus was there all along....

 

[Curious_Inquirer]

Just saw that Tech Jesus was there all along....

And Wendell! man is it great that he’s going to all these events. He’s one of the best proponents of things going Linux and open source. It’s good that AMD has invited him to see what’s going on and getting hands on time with these CPU’s, he usually has one of the deepest insights in his reviews alongside Dr. Cutress.

 

[nicalandia]

And Wendell! man is it great that he’s going to all these events.

Yes, he is the dude with Red Shirt.

 

[Curious_Inquirer]

Yes, he is the dude with Red Shirt.

Yeah I saw him, been a long time viewer of his, he inspired me to get into Linux and tinkering with software in general.

 

[IronLynx]

Greymon on Twitter about Zen4 being 30% better than Zen3 in Games: Not quite that high, but close to that number.

Wow...this chip is gonna be a beast. What Zen4XD would do?

 

[inf64]

Greymon on Twitter about Zen4 being 30% better than Zen3 in Games: Not quite that high, but close to that number.

Wow...this chip is gonna be a beast. What Zen4XD would do?

Looks like my perf. estimate will be spot on again if above is true.

 

[Saylick]

Wow...this chip is gonna be a beast. What Zen4XD would do?

Probably slap another 15% on top of the >25% gaming uplift. Gimme, gimme, gimme that 7800X3D.

 

[nicalandia]

What Zen4XD would do?

Gaming Performance on Absurd levels.

 

[Daneden]

View attachment 68059

Gaming Performance on Absurd levels.

Any source of this data available?

 

[lobz]

7-Zip HWBot doesn't look that impressive against a PBO-enabled 5950X on AIO cooling. Only about 11% faster. 108 degrees is spicy.

5950X:

7950X:

??

Ever since Zen 2 and especially Zen 3 is already exceptionally strong in decompression, 7z was never going to show that huge of an improvement. Also don't forget, there are some use cases, where there's not much advancement. For example, whatever CPU-Z is doing, AMD themselves presented basically stagnation there, and quite publicly at that...

 

[Abwx]

??

Ever since Zen 2 and especially Zen 3 is already exceptionally strong in decompression, 7z was never going to show that huge of an improvement. Also don't forget, there are some use cases, where there's not much advancement. For example, whatever CPU-Z is doing, AMD themselves presented basically stagnation there, and quite publicly at that...

AMD state 13% better IPC than Zen 3 in 7 ZIP, the 7950X score is 51% better than a stock 5950X s 151131 combined score, dunno how it can perform that well in the pic above, by comparison a 12900K does 125046.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Benchmarks in Anwendungen

Intel Alder Lake im Test: Benchmarks in Anwendungen / Ein rekordverdächtiger Testumfang / Windows 10 vs. Windows 11

www.computerbase.de

 

[tamz_msc]

??

Ever since Zen 2 and especially Zen 3 is already exceptionally strong in decompression, 7z was never going to show that huge of an improvement. Also don't forget, there are some use cases, where there's not much advancement. For example, whatever CPU-Z is doing, AMD themselves presented basically stagnation there, and quite publicly at that...

The point of that comparison was to show how crippled the 5950X is at stock TDP with 142 W PPT. When both the 7950X and 5950X are allowed to draw the maximum power possible (using normal cooling), the difference in 7zip at least is not that large.

 

[IntelUser2000]

View attachment 68059

Gaming Performance on Absurd levels.

Is that DDR4 or DDR5? Modern chips are starting to seem bandwidth limited. If Zen 4 is limited by memory bandwidth in games, then VCache will have a greater gain on Zen 4 than Zen 3 did.

 

[coercitiv]

Is that DDR4 or DDR5?

Neither, it's MLID.

 

[IronLynx]

According to APISAK Ryzen 7900X does 53009 multi (against 54513 from 13900K) in PassMark and 4382 against 4833 in single. Not even 7950! Wow!

 

[Det0x]

Some nice background information from 7min and 20sec and heron out in this video

tldr:

• CPU temp sensor goes all the way down to -49
• They could run "full pot" with their dual CCD 16core 2GMI sample -> in their words its more behaved than Zen3
• IGPU acts like Raven ridge and Picasso APU's when cold
• With LN2 mode enabled in bios you wont be voltage limited in Ryzen Master (limit is set by motherboard VRM)
• Like previous stated, auto:1:1 is recommended settings for 24/7 (MCLK+UCLK should run synced the fastest you can)
• With Zen3 you had to lower infinity fabric speed from ~1900mhz to ~1400mhz when running LN2. In this video they had lowered memory clock from ~3200mhz to ~1800mhz with their Zen4 sample
• Seems like default power limits for the 7950x is 180 TDC and 250 EDC
• "VDDG CCD" and "VDDG IOD" are mashed together into "VDDG MISC".
• Need to play with 4 main voltages when you overclock: VCORE, VSOC and VDDG MISC together with VDIMM
• They have one setup on normal ambient cooling running 2200:3200:3200 (this should be the aim for normal 24/7 settings) (6400MT/s memory speed)
• Its still easier to overclock the 7600x and 7800x compared to 7900x and 7950x with LN2 because single CCD cpus only have one GMI links compared to the dual CCD cpus which have two (GMI is still a limiting factor with Zen4 under LN2)
• The 7950x have the best silicon for cpu clockspeed (binning)

 

[CluelessOne]

AMD miss the chance to troll with their naming. Imagine if they name it 7950 and 7970 to commemorate Tahiti GCN launched 10 years ago.

 

[deasd]

According to APISAK Ryzen 7900X does 53009 multi (against 54513 from 13900K) in PassMark and 4382 against 4833 in single. Not even 7950! Wow!

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

PassMark Intel vs AMD CPU Benchmarks - High End

PassMark Software - CPU Benchmarks - Over 1 million CPUs and 1,000 models benchmarked and compared in graph form, updated daily!

www.cpubenchmark.net

some hidden message from Passmark:

7900X has 15.6% higher MT score than 5950x(45851)
13900K has 31.5% higher MT score than 12900K(41458)

Zen4 has ~10.7% higher IPC than Zen3
RaptorCove has ~6.4% higher IPC than Zen4
GoldenCove has ~2.4% higher IPC than Zen4

5950x(45851) has 16.6% higher score than 5900x(39314), so 7950x estimated score is 61823, which is 13.4% higher than 13900k.

 

[biostud]

• The 7950x have the best silicon for cpu clockspeed (binning)

Until 7800X3D hopefully

 

[Kaluan]

Until 7800X3D hopefully

Or XT refresh 🤣