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!

 

[Vope45]

Nice, AMD saved all resources for Zen 5 and their Zen 4 low end product still beat Intel top end while consuming much less power, 3dcache will have much bigger lead in games, improvement in original cache hierarchy that is optimized even more for 3dcache + second generation 3dcache will easily beat RPL and reclaim gaming crown, and Zen 5 is completely new architecture, AMD have bright future.

I don't recall amd saving all resources for zen 5.

 

[CakeMonster]

Disagree about cache being optimized even more for V-Cache, that's a lucky coincidence if its the case, its naive to think Z4 was designed primarily for the lower volume 3D parts than for the main lineup.

Z5 was hinted at being good by that AMD dude in Ian's interview a while back, but that's all we have to go on. I'm very much inclined to believe it, but you're inferring way too much here.

 

[FangBLade]

Disagree about cache being optimized even more for V-Cache, that's a lucky coincidence if its the case, its naive to think Z4 was designed primarily for the lower volume 3D parts than for the main lineup.

Z5 was hinted at being good by that AMD dude in Ian's interview a while back, but that's all we have to go on. I'm very much inclined to believe it, but you're inferring way too much here.

Zen 4 isn't designed primarily for 3dcache, but it is better optimized for it based on feedback from previous architecture, they changed/improved whole cache hierarchy, and double l2cache will help 3dcache, also original l3 have upto 50% more bandwith, all that will help with cache misses and latency, also design of whole architecture is tweaked to have lower heat when combined with 3dcache, 3dcache is their trump card , first generation was beta testing product to see how it translates in desktop workload, it was successful, and they will use it more and more, also improving it, and as yields grow more products will use it. About resources, small team worked on Zen 4, while everyone else focused on Zen 5, it was same as with Zen+ and Zen 2, AMD even confirmed that.

 

[Timmah!]

I don't recall amd saving all resources for zen 5.

Yeah, this. Its ok to be excited for next-gen products, but this is bit over the top. Zen 5 might turn out great and it probably will, but no need to wax lyrically how uber-awesome its gonna be - cause it wont, regardless of what AMD has been hyping. Lets be honest, things are very rarely too good to be true, there is never a free lunch and almost always some caveat. With Zen4, despite all the great things about it, its increased power consumption and possibly temp issues and you can bet there will be some downsides to Zen5 as well. If not performance or power-draw, then it will be price. You can bet on that.

 

[fkoehler]

Much as I strongly prefer AMD, I think Charlie is right to harp on them for implementing Pluton or whatever MS is trying to get baked in across the industry.

Semi-Accurate is on my daily checklist, and most of his complaints about AMD's Marketing seem to be pretty fair game.
It appears that Lisa is on the up and up, perhaps helping to sandbag as necessary, however Marketing is playing stupid games.

He's not much different than usual, he's just been turning on AMD more and more in recent times so they're getting a bit of his typical Intel/Nvidia treatment.

 

[Kaluan]

Greymon55's comments on one of the first (weak MT) 7950X CB leaks we got:

https://twitter.com/x/status/1566021507574562816

 

[uzzi38]

Why ? Repeat after me,

Moving data hard , executing data easy.

Remember this core is smashing a core that has way more decode, rob, load store, l1d and execution units / prf ports.

Zen4 is frankly amazing probably the best zen core since zen1.

So much performance for so little functional resource increase. Zen5 is gonna be stonks!!!!!

Maybe not the best expectation for Zen 5.

 

[uzzi38]

I thought AMD had said that there would be a Zen X3D product out before the end of the year? Have they changed their plans?

Information wasn't really provided well by AMD. They suggested that with their wording, but I don't see any chance we see X3D this year personally.

 

[Abwx]

Information wasn't really provided well by AMD. They suggested that with their wording, but I don't see any chance we see X3D this year personally.

August 29 lisa Su said "later this year" for the X3D variants.

 

[maddie]

August 29 lisa Su said "later this year" for the X3D variants.

He did say personally.

 

[Joe NYC]

Much as I strongly prefer AMD, I think Charlie is right to harp on them for implementing Pluton or whatever MS is trying to get baked in across the industry.

On Pluton, I agree.

To add insult to injury (injury of AMD customers by exposing them to government surveillance and mistreatment), Microsoft adds an insult of not using AMD's superior CPU - Rembrandt - for Surface.

 

[uzzi38]

He did say personally.

Yeah, honestly I'd like to be wrong, but my expectation is more H1 next year rather than late this year.

 

[DrMrLordX]

Yeah, honestly I'd like to be wrong, but my expectation is more H1 next year rather than late this year.

Same. If they have Raphael-X parts out too soon, it may hurt sales of standard Raphael. Need to give it a few months to sell through and recoup development costs (which admittedly are mostly being paid for by Genoa).

 

[Kaluan]

Greymon55 again:

https://twitter.com/x/status/1566761996355530752

Newer/better BIOS/microcode?

 

[IEC]

I wouldn't be shocked if AMD is sandbagging. They have relatively consistently underpromised and overdelivered since Lisa Su took the helm.

I expect Zen 4 and Raptor Lake to be within spitting distance of each other for most tasks. However, I don't expect Intel will have an answer for Zen 4 3D before 2024 which is why I am waiting for that launch to decide my next system.

 

[Kaluan]

Greymon55 again:

https://twitter.com/x/status/1566761996355530752

Newer/better BIOS/microcode?

Seems Harukaze is hinting at something similar, but related to drivers and gaming performance:

https://twitter.com/x/status/1566730432955052033

Edit: In Gremon55's BiliBili video link, I think that person mentions ~59ns is possible at 6400CL32 (2133:3200:3200), I don't speak Mandarin tho, if someone can confirm that'd be great.

 

[Timmah!]

Greymon55 again:

https://twitter.com/x/status/1566761996355530752

Newer/better BIOS/microcode?

82C stock or that OC?
Anyway, sounds good.

 

[randomhero]

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-4-genoa-ryzen-7000-etc.2571425/post-40737983

Ha! Guess I was right!

For all THE WRONG reasons!

If you have ever been in a situation where you had to resolve complicated math function and you screw up starting parameters and algebra too while calculating, but got correct final result, than this is it.

 

[randomhero]

I have to say that I am impressed with Zen4. I am looking forward for Phoenix and Dragon disclosure at CES.
Those chips will be something else,my 5900HX will look like is from dark ages.

 

[Kedas]

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-4-genoa-ryzen-7000-etc.2571425/post-40737983

Ha! Guess I was right!

For all THE WRONG reasons!

You have a lesson learned, never explain yourself and you will be fine

 

[coercitiv]

Always be suspicious of people around here who predict things without explaining.

They either know too much, or too little

 

[rtxtwt]

Greymon55 again:

https://twitter.com/x/status/1566761996355530752

Newer/better BIOS/microcode?

Badly made fake from china bilibili, some accounts are just repeat doing things like this, and "leakers" like greymon couldn't figure out things about AMD,,,,,

https://twitter.com/x/status/1566818410381553665

 

[Mopetar]

If you have ever been in a situation where you had to resolve complicated math function and you screw up starting parameters and algebra too while calculating, but got correct final result, than this is it.

So are you telling me that two wrongs does make a right?

 

[AtenRa]

If true this is huge

AMD Ryzen 5000 CPUs See Major Price Cuts As Ryzen 7000 Retail Launch Approaches: 5800X3D Down To $379, 5950X at $499, 5900X at $349, 5800X at $239, 5600X at $179

With the retail launch of the Ryzen 7000 approaching, the existing AMD Ryzen 5000 CPUs have started to receive some pretty major price cuts.

wccftech.com

 

[Mopetar]

Might not be anything official from AMD and retailers tend to discount older generation parts before a launch, but it doesn't matter what the reason is if you can get a cheaper part.

Zen 3 is still a great CPU and even though AM4 is EOL if the price is cheap enough, it can make a great system.