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!

 

[deasd]

I love the smell of astroturfing in the morning

I just remembered when old days some people use undervolting AMD Thuban X6 10xxT to show strong efficiency against the Nehalem i7......man thats nostalgic. Although AMD suffered at that time but still nostalgic and funny.

 

[Det0x]

Haven't seen anything posted about this in this thread:

Robert Hallock to Leave AMD After 12 Years

Robert Hallock, AMD's director of technical marketing, is departing the company after 12 years, he announced on LinkedIn(opens in new tab).

Hallock was a public-facing figure at the company, working with the AMD fan community on Reddit, Discord, YouTube and other platforms while also briefing the press.

"Over the years, I've had the honor and privilege of publicly teaching others about some truly stellar innovation: the Zen core family, 3D V-Cache, chiplet packaging, HBM memory, FreeSync, low-overhead graphics APIs, and much more," Hallock wrote in his post. "After working in both graphics and processors for roughly 6 years each, I've learned so much."

Hallock hasn't detailed where he'll be going next. On LinkedIn, he wrote that he would be traveling while he thinks about his next move. Hallock is one of several high-profile employees at AMD to leave in recent years.

I think this could be a good thing in the long run as i haven't been especially impressed by the marketing lately..
Just one example, what do you guys think sound better: (recent blender comparison with 12900k vs 7950x)

• "31% faster" (its actually 31% less time)

or

• "46% faster" ?

https://twitter.com/x/status/1528756387781693440

 

[moinmoin]

Haven't seen anything posted about this in this thread:

Robert Hallock to Leave AMD After 12 Years

A personable guy. Could have been much more pedantic with numbers though indeed.

 

[naad]

A lot of these power comparisons don't take into account the static 20W AMD pulls due to the I/O die, making it a good 30-50% of total draw at low TDP comparisons while Intel's ring goes down to 2-5W

Real (core to core) comparisons would be with AMD's APUs, but those have reduced caches so they trade some IPC for power efficiency, and that's why you see 6800U mow down ADL mobile when uncore isn't tying the cores down.

Also, use of something else than cinnebench for these comparisons would be preferable.
Hopefully Phoenix's L3 cache isn't cut down so we can get a real comparison with all other variables equal.

 

[Kocicak]

OG Zen, up to Zen 3 are not that efficient, thanks to the IOD, in single and few cores scenarios.

The monolithic G Ryzens exist, you know... I recently tested efficiency of 1 thread on this CPU and it is excellent compared to 12600K.

Page 3 - Discussion - New computer parts as an enviromental disaster?

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[DisEnchantment]

Hopefully Phoenix's L3 cache isn't cut down so we can get a real comparison with all other variables equal.

I think Phoenix will be a good show case for x86 CPUs. RDNA3 will shine if rumors about focus on area and power efficiency are true.
Zen4 is also quite area efficient, on top of that the N4 node should give it a slight efficiency edge vs vanilla desktop Zen4. AIE could be interesting but probably not as interesting as it would have been on cellphone SoCs.

Next two years of Mobile products from AMD will be formidable,
Phoenix --> Zen4+RDNA3+AIE on N4.
Strix Point --> Zen5 + RDNA3+ and AIE on N3
It does seem AMD will remain unchallenged in terms of efficiency in x86 mobile space.

I think the burgeoning Handheld gaming devices market will explode if RDNA3 delivers big on efficiency and performance on top of the massive gains in overall SoC efficiency due to N4.

 

[Hans Gruber]

I think Phoenix will be a good show case for x86 CPUs. RDNA3 will shine if rumors about focus on area and power efficiency are true.
Zen4 is also quite area efficient, on top of that the N4 node should give it a slight efficiency edge vs vanilla desktop Zen4. AIE could be interesting but probably not as interesting as it would have been on cellphone SoCs.

Next two years of Mobile products from AMD will be formidable,
Phoenix --> Zen4+RDNA3+AIE on N4.
Strix Point --> Zen5 + RDNA3+ and AIE on N3
It does seem AMD will remain unchallenged in terms of efficiency in x86 mobile space.

I think the burgeoning Handheld gaming devices market will explode if RDNA3 delivers big on efficiency and performance on top of the massive gains in overall SoC efficiency due to N4.

I am just going to throw it out there. Nvidia does a much better job of implementing drivers that get the most out of games/hardware. In a year from now, it could be possible the RX6800 could equal a 3080 simply because AMD issues big driver updates with real substantial performance gains. The problem, they do the major driver updates every 6 months.

So when people read 30% improvement on Game X or 25% improvement on Game Y. It's because their drivers are not optimized.

Nvidia releases new drivers every week. I honestly do not think they can improve on the current generation of Nvidia cards. Sometimes they release bad drivers but the next Nvidia driver comes out the following week.

The other thing. Cinebench R23 all core is not realistic real world conditions. It's like Prime 95 light. I was testing my Zen 3 CPU last night on BF2042. The watts used gaming was between 60-71w while gaming. The vast majority of the time the 5600 was pulling 64-68w.

So expect Zen 4 CPU's to pull close to what the TDP is. 105w should be about 105w while playing a AAA game title.

I expect the RDNA 3 GPU's to be limited by their drivers. So performance will be lower and could get much better over time.

 

[Zepp]

I think Phoenix will be a good show case for x86 CPUs. RDNA3 will shine if rumors about focus on area and power efficiency are true.
Zen4 is also quite area efficient, on top of that the N4 node should give it a slight efficiency edge vs vanilla desktop Zen4. AIE could be interesting but probably not as interesting as it would have been on cellphone SoCs.

Next two years of Mobile products from AMD will be formidable,
Phoenix --> Zen4+RDNA3+AIE on N4.
Strix Point --> Zen5 + RDNA3+ and AIE on N3
It does seem AMD will remain unchallenged in terms of efficiency in x86 mobile space.

there is rumor that Phoenix will use RDNA2

Page 13 - Question - AMD Phoenix/Zen 4 APU Speculation and Discussion

Page 13 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[nicalandia]

Look what I found on my daily morning routine

Details AMD Ryzen 5 7600X 6-Core Processor 6C 12T 4.7GHz

Details for Result ID AMD Ryzen 5 7600X 6-Core Processor (6C 12T 4.7GHz, 1.33GHz IMC, 6x 1Mo L2, 32Mo L3)

ranker.sisoftware.co.uk

 

[jpiniero]

Look what I found on my daily morning routine

Details AMD Ryzen 5 7600X 6-Core Processor 6C 12T 4.7GHz

Details for Result ID AMD Ryzen 5 7600X 6-Core Processor (6C 12T 4.7GHz, 1.33GHz IMC, 6x 1Mo L2, 32Mo L3)

ranker.sisoftware.co.uk

A quick look and that's about what the average 12600K gets.

 

[nicalandia]

A quick look and that's about what the average 12600K gets.

Yes, indeed which is impressive since the 12600K is 6C/12T P cores + 4 e-cores



Those e crores are giving a 40% boost to the 6C/12T CPU in Int performance, but not much in FP but it's known that e cores have weak FPU

P Core vs P Core the 7600X just steam rolls the Golden Cove CPUs


Also the 7600X its within 6% of the 5800X performance.

 

[deasd]

Sandra Arithmetric Native utilize AVX and AVX2, no AVX512 being utilized(look at the i9-7900X which has native avx512 support)

AVX2 for integer, AVX for FP




so 7600X has 10% less FP but comparable INT perf to 5800X. Zen4 is Integer-focus architecture.

 

[Mopetar]

For the very first time in the last decade AMD CPUs will have higher rated max boost than Intel?

I suppose it's kind of nifty, even if it's only for a month or so.

 

[uzzi38]

Anyways, the 13900T will be the most efficiency cpu on planet, followed by the 13700t, out of the box, hands down.

If you want this statement to hold true, I highly recommend you add some limits to the statement, like saying the most efficient desktop CPU out of the box that you'll be able to buy at retail. Also maybe putting a time restriction on the statement, like saying "for the next year".

 

[nicalandia]

I suppose it's kind of nifty, even if it's only for a month or so.

The 13700K will have a 5.4 Ghz P-core Max frequency and the 13900KS will have a 5.7 Ghz P-core Max frequency. and 6 Ghz with Thermal Boost Velocity.

Do you count Thermal Boost Velocity as Max boost clocks?

 

[uzzi38]

there is rumor that Phoenix will use RDNA2

Page 13 - Question - AMD Phoenix/Zen 4 APU Speculation and Discussion

Page 13 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

It's officially announced as RDNA3. That rumour can be put to rest (and was a while ago already...)

 

[tamz_msc]

The 13900KS will have a 5.7 Ghz P-core Max frequency. and 6 Ghz with Thermal Boost Velocity.

Most certainly not. TVB frequency is just one bin higher than TBM frequency (see below post), and both are higher than P-core Max frequency, which is for all core loads. There is no 13900KS yet, so it is meaningless to speculate what frequency it will have.

 

[uzzi38]

Most certainly not. TVB frequency is just one bin higher than TBM frequency

We've seen TVB up to two stages already. Although I doubt the idea that it'll be up to 3 stages, this statement isn't actually true.

 

[tamz_msc]

We've seen TVB up to two stages already. Although I doubt the idea that it'll be up to 3 stages, this statement isn't actually true.

Where does TVB go two bins or more over and above TBM?

Edit: It looks like the 12900KS is actually having TVB two bins above TBM. My bad.

 

[nicalandia]

There is no 13900KS yet, so it is meaningless to speculate what frequency it will have.

Intel Teased a 6Ghz clock recently. Also there is a 13900KS that was OC to the moon.

So that has to be the 13900KS and it will be a safe bet that it's not their P-Core Max boost, but likely the TVB. If that is the case can you extrapolate the All P-Core from that number?

 

[tamz_msc]

Intel Teased a 6Ghz clock recently. So that has to be the 13900KS and it will be a safe bet that it's not their P-Core Max boost, but likely the TVB. If that is the case can you extrapolate the All P-Core from that number?

Intel hasn't disclosed just how it plans to market 6 GHz, except to say that it is coming with Raptor Lake. It is likely to feature as a -KS model, but it is meaningless to speculate about which turbo bin will get a particular label.

 

[Hans de Vries]

Yes, indeed which is impressive since the 12600K is 6C/12T P cores + 4 e-cores

Visualized:


Higher efficiency, higher base clock, maybe AVX512 although that's running at half speed.

 

[nicalandia]

Visualized:

Higher efficiency, higher base clock, maybe AVX512 although that's running at half speed.

Thanks for sharing.

In Sisoftware the AVX-512 is measured by the CPU Quad-Float performance. The 7600X was not tested yet on Processor Multi-Media. So we don't have numbers on that yet.

 

[Mopetar]

The 13700K will have a 5.4 Ghz P-core Max frequency and the 13900KS will have a 5.7 Ghz P-core Max frequency. and 6 Ghz with Thermal Boost Velocity.

Do you count Thermal Boost Velocity as Max boost clocks?

I haven't been paying attention to whatever fancy sounding names companies are using, but if it will go to 6 GHz out of the box without needing me to dig around in the BIOS too terribly much then yeah, I'd consider it as a max boost unless it's only possible with some kind of exotic cooling solution.

Even if it doesn't meet those criteria, how long until Intel has a 13900KF that they've pushed that little bit further just so it can be king with a 5.8 GHz max frequency?

Frankly though I wouldn't want to run it that high. The extra ~5% performance is probably going to translate into something like ~50% more power and heat. If I did buy a Raptor Lake CPU then I'd probably try to tune it for better efficiency even if it meant dropping the boost duration or limit.

 

[nicalandia]

Even if it doesn't meet those criteria, how long until Intel has a 13900KF that they've pushed that little bit further just so it can be king with a 5.8 GHz max frequency?

The 13900KF is not faster than the standard 13900K we have plenty of geekbench entries to support that. It's just a 13900K without iGPU.