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!

 

[moinmoin]

I think the architecture should come before the product category or maybe even before the year. It prevents something like a customer thinking an 9820HX is better than a 8755HX, though I doubt AMD would release something like this extreme example.

I agree but alas as we know, for OEMs and marketing...
Second best choice would have been re-using the server model numbering system of putting the architecture as last number.
But having architecture be a part of the model numbering system at all is a huge improvement for AMD mobile systems which were a mess up to now.

 

[Joe NYC]

All possibly true but still bit unfortunate. If Zen5 is coming 2024, i would prefer to build Zen4 rig still in 2022, not 2023, to avoid buyers remorse, especially if Zen5 will be significant improvement, as rumored.

I will believe those pricedrops when i see them

If the time span between Zen 4 and Zen 5 is a year and a half, I think the time gap between Zen 4 3D and Zen 5 will be at least a year. Or more...

 

[gdansk]

In isolation that is an OK naming scheme. But in context of previous naming schemes, isn't it a bit exploitative? The most significant digit was the generation. Well I guess they already did that with the 5700U and at least now there is a number to indicate it is last gen core.

 

[igor_kavinski]

That's actually a great change if they consistently stick to it.

Someone must be writing an online AMD CPU model decoder right now...

 

[moinmoin]

The most significant digit was the generation.

And that was a mess in mobile from the very beginning. Was 2000 Zen+? No it was Zen. Was 3000 Zen 2? No it was Zen+. 4000 finally cleaned it up until 5000 made it a mix of Zen 2 and 3, great! That's why having the architecture generation as a specific digit would/will be such an big improvement.

 

[DeathReborn]

Someone must be writing an online AMD CPU model decoder right now...

It seems AMD already has offline decoders, and they are "special" shall we say...

 

[LightningZ71]

I almost don't mind the rebranding of Rembrandt given that there have been so few models hitting the market, it's likely to not really confuse anyone.

 

[RnR_au]

That's actually a great change if they consistently stick to it.

Disagree.

The 'year' has no place being at the front. And why a '7' for 2022? Clearly to suggest some connection to Ryzen 7000 on the desktop. Its just marketing confusion for the masses.

Thankfully I don't have hordes of family members who buys computing gear before asking me 'why is this...' so the change won't affect me. Just saddened that marketing has won over engineering at AMD.

 

[itsmydamnation]

Disagree.

The 'year' has no place being at the front. And why a '7' for 2022? Clearly to suggest some connection to Ryzen 7000 on the desktop. Its just marketing confusion for the masses.

Thankfully I don't have hordes of family members who buys computing gear before asking me 'why is this...' so the change won't affect me. Just saddened that marketing has won over engineering at AMD.

its obviously because 2022 is divisible by 7.... like DER

Spoiler

yes, i am not that stupid

Spoiler

/S

 

[moinmoin]

The 'year' has no place being at the front. And why a '7' for 2022? Clearly to suggest some connection to Ryzen 7000 on the desktop. Its just marketing confusion for the masses.

I think to a degree it's just honest, AMD does yearly APU launches and OEMs want yearly refreshes on top of that. 6000 is already in use and 7000 will be for 2023.

I'm actually interested in what AMD intends to do with model numbering on AM5. On desktop they already skipped 4000 so they could (and imo ideally should) use the same system, at least including the architecture generation digit. They'll likely keep skipping most significant digits that were already launched as mobile anyway.

 

[jpiniero]

I think to a degree it's just honest, AMD does yearly APU launches and OEMs want yearly refreshes on top of that. 6000 is already in use and 7000 will be for 2023.

The issue is that they are refreshing old products.

I'm actually interested in what AMD intends to do with model numbering on AM5.

Doesn't seem like they are changing the scheme on desktop.

 

[NostaSeronx]

Bulldozer --> Zen took 5+ years.

Going based by production, kind of ignores how absurd Bulldozer actually was.

Bulldozer's development started after K5. (March 1996+)
Zen's development started after Bobcat. (January 2011+)

Bulldozer went through several CTOs/Chief Architects each with different visions, while Zen stayed within a single vision.
(Late 90s/Early 00s/Late 00s each had different targeting points for each Bulldozer design. 2007+ was bum rushing BD into Tablets(Mobility), not into Servers(Computing), while also not doing that.;;
Bulldozer 45nm/32nm (before mid-2007: no microsoft involvement) => Cluster-based Multithreading architecture on automated methodology => High-IPC HPC focus
Bulldozer 32nm (after mid-2007: microsoft co-architect involved in CPU architecture design for never released Xbox console) => Chip-level Multithreading architecture on mostly-hand methodology => High-Frequency Client focus)

2002+ = Bulldozer ditches mixed-point cluster (Integer + Floating Point within same cluster::bad because AMD also developed grid-style execution: SIMD-units for superscalar execution::128-bit datapath can be used for two independent 64-bit ops or four independent 32-bit ops:, sometime in 2003-2005 which they ditched.)
2007+ = Bulldozer ditches cluster-based multithreading(2Cluster-1Thread mode as well as 2C-2T mode) for chip-level multithreading(2Core-2Thread mode only).

1997 (First Bulldozer architecture work start) -> 2010/2011 (First Bulldozer product showing/release) = ~13 years
2012 (Zen architecture work start) -> 2016/2017 (First Zen product showing/release) = ~5 years

The amount of development time within the Zen era would be the same as the Bulldozer era by 2025.

 

[Thunder 57]

It seems AMD already has offline decoders, and they are "special" shall we say...

I don't know who the hell thought this crap up, by holy hell. You need a decoder to figure out what's what? I don't like marketing people in general. This is just an over the top example of how stupid one can be. I thought "GameCache" was bad. This is orders of magnitude worse.

 

[deasd]

7700X CPUZ, ST 774 MT 8381, original link was deleted, voltage and temperature(71.6c mentioned by comment) are good

https://twitter.com/x/status/1567684911645204480

 

[Det0x]

Want to see some gaming benchmarks, but here are some numbers in CPU-Z and GB5 from the leak above:

AMD Ryzen 7 7700X perforamnce in CPU-Z and Geekbench has been leaked

Starting off with CPU-z, the AMD Ryzen 7 7700X was running on the Gigabyte X670E AORUS Master motherboard and this is the retail variant. The CPU was running at 5.42 GHz all-core boost clocks & had an operating voltage of 1.152V with temperatures hovering around 71C. We don't know if this was under load or what type of cooling was used. The maximum multiplier reported for the chip is 55.5x which leads to a 5.5 GHz peak frequency limit.




With such a score, the CPU is 21% faster than Ryzen 7 5800X (8-core Zen3) and 20% faster than Ryzen 9 5950X (16-core Zen3) in single-core test. It is also 1% faster than Core i5-12600K (10-core Alder Lake), but it is outperformed by Core i7-12700K (12-core).

In terms of multi-threaded performance, the 7700X scores 1% higher than Core i9-12900K (16-core) and 8% higher than i7-12700K. This new Raphael CPU is also 28% faster than its predecessor (5800X).

There is also this new entry to Geekbench database with the same CPU but different motherboard (ASUS ROG Crosshair X670E Hero) and memory configuration (64GB of DDR5-6000):



Here, the CPU scores 2209 and 14459 points in single and multi-core Geekbench V5 tests respectively. In this case, 7700X ends up faster than Core i7-12700K by 16% in single-core and 2% in multi-core tests.

AMD Ryzen 7 7700X is officially set to launch on September 27th alongside three other Ryzen 7000 CPUs. According to the reports, AMD will lift the review embargo Ryzen 7000 as well as X670 motherboards a day sooner (September 26th).

 

[bakyt115]

In terms of multi-threaded performance, the 7700X scores 1% higher than Core i9-12900K (16-core) and 8% higher than i7-12700K

now time to slash zen mt score.

 

[gdansk]

the 7700X scores 1% higher than Core i9-12900K (16-core)

With 16 threads. But what does 12900K score with all 24 threads? Around 11k

 

[deasd]

With 16 threads. But what does 12900K score with all 24 threads? Around 11k

yeah VCZ just compare 8C16T of every SKUs from both side, but still never expected they perform that similar with same core/thread count, since Zen4 CPUZ IPC has no uplift.

this 7700X above operated at 5.4Ghz all core and temperature is good, we don't know the cooling method but I guess with this voltage I expect 360AIO at most.

 

[Just Benching]

Want to see some gaming benchmarks, but here are some numbers in CPU-Z and GB5 from the leak above:

AMD Ryzen 7 7700X perforamnce in CPU-Z and Geekbench has been leaked

Starting off with CPU-z, the AMD Ryzen 7 7700X was running on the Gigabyte X670E AORUS Master motherboard and this is the retail variant. The CPU was running at 5.42 GHz all-core boost clocks & had an operating voltage of 1.152V with temperatures hovering around 71C. We don't know if this was under load or what type of cooling was used. The maximum multiplier reported for the chip is 55.5x which leads to a 5.5 GHz peak frequency limit.
View attachment 67283

View attachment 67284

With such a score, the CPU is 21% faster than Ryzen 7 5800X (8-core Zen3) and 20% faster than Ryzen 9 5950X (16-core Zen3) in single-core test. It is also 1% faster than Core i5-12600K (10-core Alder Lake), but it is outperformed by Core i7-12700K (12-core).

In terms of multi-threaded performance, the 7700X scores 1% higher than Core i9-12900K (16-core) and 8% higher than i7-12700K. This new Raphael CPU is also 28% faster than its predecessor (5800X).

There is also this new entry to Geekbench database with the same CPU but different motherboard (ASUS ROG Crosshair X670E Hero) and memory configuration (64GB of DDR5-6000):



Here, the CPU scores 2209 and 14459 points in single and multi-core Geekbench V5 tests respectively. In this case, 7700X ends up faster than Core i7-12700K by 16% in single-core and 2% in multi-core tests.

AMD Ryzen 7 7700X is officially set to launch on September 27th alongside three other Ryzen 7000 CPUs. According to the reports, AMD will lift the review embargo Ryzen 7000 as well as X670 motherboards a day sooner (September 26th).

I think it must be beta bios. How can it lose to the 12700k in both st and multi?? Thats absurd.

 

[igor_kavinski]

Bear in mind you can have FiVR while also having DLVR should Intel and it's partners wish.

Do the 50 new power management features in Zen3+ include FiVR and DLVR? If not, isn't it impressive what they achieved and what they can if they now put these two features in Zen 4?

 

[exquisitechar]

I think it must be beta bios. How can it lose to the 12700k in both st and multi?? Thats absurd.

“IPC” improvement is 1% in CPU-Z according to AMD, so no surprise there.

 

[inf64]

CPUz "benchmark", just laughable. Let's see some real benchmarks that test productivity and do real work instead of these junk tests.

 

[Asterox]

CPUz "benchmark", just laughable. Let's see some real benchmarks that test productivity and do real work instead of these junk tests.

A little more specific, he should be sent to the Russian gulag.

 

[moinmoin]

The issue is that they are refreshing old products.

Indeed, nothing changed in that regard. And while that happens it's better to be upfront about it within the existing model numbers than hide that fact away with some separate inconsistent model numbering rules as happened in the past gens.

Doesn't seem like they are changing the scheme on desktop.

That's highly unfortunate since some mobile chips are bound to come to AM5 later, and the same transparency won't be there then.

 

[nicalandia]

16T vs 16T 7700X vs 12900K


Some dude in Japan.