Speculation: Ryzen 4000 series/Zen 3

[Thunder 57]

As a matter of fact, AMD showed SMT2 in the presentation. The trick is that SMT4 core can be switched into SMT2 mode too, so that presentation was correct. AMD never said Zen3 has no SMT4. Prove me wrong and find any official denial of SMT4 by AMD.

Tachyum Prodigy is resurecting code-morphing Transmeta-like core. What if Zen3 uses code-morphing too? Maybe K12 was not canceled but rather merged with Zen3. So Zen3 will be able to run x86 code as well as ARM binaries. And support both AVX and SVE2 SIMD. AMD could gain from ARM's new features and faster development (better ISA ARMv9, 6xALUs, 2048-bit SVE2 FPU) while keeping x86 backward compatibility. That would be pretty smart move.

I have proven you wrong before. There will be no SMT4 in Zen 3. You are a fool to believe otherwise.

 

[uzzi38]

Richie, your theory about Zen3 and K12 being related is - as it so happens - wrong. You're not far off from a similar truth however, but you're too fixated on Zen 3 and you're not thinking logically as to what would actually be closest to K12.

And for the record, K12 is very definitely canned. Resurrectable if needed in the future, but for now canned. Honestly, if AMD actually do resurrect it and bring it up to speed with the 2020s, it would be REALLY interesting.

But for now, it's dead.

As for SMT4, can we please stop beating the poor, poor dead horse?

 

[Richie Rich]

You've proven that SMT4 Zen3 will have SMT2 mode too. Nothing else as far as I remember. Even Mark Papermaster at interview didn't deny SMT4 when redactor asked direct question about Zen3 and SMT4. I wonder why he didn't say "there is no SMT4 in Zen3 as you saw in the presentation slide back in August 19". He could say that without any damage but he didn't. He let the SMT4 possibility open for some reason....

Some people are fighting against such a cool feature as SMT4 so hard that it looks like they are AMD employees trying to keep it secret until Zen3's launch day

BTW: The K12 and Zen3 code-morphing core was meant as joke/wild speculation even it would be possible in theory. These days it's pretty hard to encourage people to turn on and use the most powerfull graphic card on the world - human imagination. It looks like everybody would be happy with 4xALU, SMT2 CPU and 3% yearly IPC increase forever. No wonder a lot of people has such a hard time with Apple's 6xALU monster core like A13. We'll se in two months if Zen3 is SMT4 or not

 

[Thunder 57]

You've proven that SMT4 Zen3 will have SMT2 mode too. Nothing else as far as I remember. Even Mark Papermaster at interview didn't deny SMT4 when redactor asked direct question about Zen3 and SMT4. I wonder why he didn't say "there is no SMT4 in Zen3 as you saw in the presentation slide back in August 19". He could say that without any damage but he didn't. He let the SMT4 possibility open for some reason....

Some people are fighting against such a cool feature as SMT4 so hard that it looks like they are AMD employees trying to keep it secret until Zen3's launch day

BTW: The K12 and Zen3 code-morphing core was meant as joke/wild speculation even it would be possible in theory. These days it's pretty hard to encourage people to turn on and use the most powerfull graphic card on the world - human imagination. It looks like everybody would be happy with 4xALU, SMT2 CPU and 3% yearly IPC increase forever. No wonder a lot of people has such a hard time with Apple's 6xALU monster core like A13. We'll se in two months if Zen3 is SMT4 or not

You are being foolish. There is no SMT4. I have given you proof. Yet you still run your mouth. Shut up and wait for it to come out. Keeping it a secret? What a joke! 6 ALU's, shut up. You don't think Intel or AMD could come up with a 6 ALU design? Yes, we will soon see if Zen 3 has SMT4 or not. When you are proven wrong, will take a slice of humble pie?

 

[gdansk]

At this point Zen3 SMT4 discussion resembles something of a theological debate. But in this case we'll see the truth in a few months so I'm not sure it's worth continuing the subject.

 

[Makaveli]

At this point Zen3 SMT4 discussion resembles something of a theological debate. But in this case we'll see the truth in a few months so I'm not sure it's worth continuing the subject.

What I want to see is if Richie will admit he was wrong once the product is out

 

[Olikan]

LOL, SMT-4 rumour is indeed alive... From Latest Momomo_us leak

 

[RetroZombie]

I wonder why he didn't say "there is no SMT4 in Zen3 as you saw in the presentation slide back in August 19".

Imagine Mark Papermaster talking about everything that isn't inside the cpu, it would be one hell of presentation

 

[RetroZombie]

LOL, SMT-4 rumour is indeed alive... From Latest Momomo_us leak

Again? I already said it's the number of sockets for sure.
The * is a warning that the socket for the 2X cpu doesn't work with a 4X cpu.

Edit: No zen4? Two zen3 releases?

 

[Olikan]

Edit: No zen4? Two zen3 releases?

Yes, no Zen 4... this back up the wierd "zen3+" rumour...
Also, There is a (Max core/threads) in the left corner now... maybe only some Milan server will have smt-4

Genoa seems like a fast upgrade, a have a new IO die, SMT-4... time to market maybe?

 

[Richie Rich]

When Zen3 will not have SMT4 I will say: "OK. I was wrong, you we right. But it's a missed opportunity to be more advanced over Intel."

Especially funny if Golden Cove in Alder Lake next year will bring SMT4. I bet all AMD fans will turn suddenly 180° and will call for SMT4 ASAP

I don't understand why some people are risking brain stroke always when I mention SMT4. Just for different opinion? That's silly. OTOH I understand 6xALU brain stroke. It's pretty embarrassing that powerfull desktop CPUs have only 4xALU. I hope Zen3 will change that and will have 6xALU back end.

 

[Thunder 57]

When Zen3 will not have SMT4 I will say: "OK. I was wrong, you we right. But it's a missed opportunity to be more advanced over Intel."

Especially funny if Golden Cove in Alder Lake next year will bring SMT4. I bet all AMD fans will turn suddenly 180° and will call for SMT4 ASAP

I don't understand why some people are risking brain stroke always when I mention SMT4. Just for different opinion? That's silly. OTOH I understand 6xALU brain stroke. It's pretty embarrassing that powerfull desktop CPUs have only 4xALU. I hope Zen3 will change that and will have 6xALU back end.

I guess I'll have a stroke then. You are wrong and you should know it. Good news is when Zen 3 comes out I will not have a stroke. Bad news for you is you will have to eat a can of cat food . And what is with your ALU dick measuring?

 

[dr1337]

LOL, SMT-4 rumour is indeed alive... From Latest Momomo_us leak

This is a glaringly obvious photoshop/recolor of the HPC slide that was released last year, the biggest giveaway is that genoa is still in definition despite this slide being almost a year newer.



Also I checked momomo's twitter and haven't seen this new slide anywhere, and reverse image search also comes up with nothing. And its very much unlike AMD to ever re-use marketing materials in such an obvious way. And the little text at the bottom that says "NDA required" to me just is screaming Obvious Troll.

Feel free to source/prove me wrong otherwise but the "new" SMT4 slide is a fresh hot fake

 

[Olikan]

This is a glaringly obvious photoshop/recolor of the HPC slide that was released last year, the biggest giveaway is that genoa is still in definition despite this slide being almost a year newer.

View attachment 24282

Also I checked momomo's twitter and haven't seen this new slide anywhere, and reverse image search also comes up with nothing. And its very much unlike AMD to ever re-use marketing materials in such an obvious way. And the little text at the bottom that says "NDA required" to me just is screaming Obvious Troll.

Feel free to source/prove me wrong otherwise but the "new" SMT4 slide is a fresh hot fake

https://twitter.com/x/status/1276906886583869443

 

[Ajay]

https://twitter.com/x/status/1276906886583869443

And just below that post, an updated release from AMD:

 

[KOMACHI_ENSAKA]

It's real slide. It's from 2018Q4. That roadmap has been in existence since before.
even before the current roadmap was released at HPC AI Advisory Counsil's 2019 UK Conference.
It is old and different from the current roadmap, but it is not a fake.

 

[uzzi38]

It's real slide. It's from 2018Q4. That roadmap has been in existence since before.
even before the current roadmap was released at HPC AI Advisory Counsil's 2019 UK Conference.
It is old and different from the current roadmap, but it is not a fake.

Oh, I'd heard about SMT4 being on the roadmap before, but I didn't realise this was an actual slide from AMD. Huh, good to know.

 

[amd6502]

Oh, I'd heard about SMT4 being on the roadmap before, but I didn't realise this was an actual slide from AMD. Huh, good to know.

Makes a lot of sense because it's also called Zen4.

On the same slide, https://img.alicdn.com/imgextra/i3/169721066/O1CN01ERGFnv1XDfBU3rCBT_!!2543522890.png we can see that 64/2X for Zen3/Milan has a superscript *.

I couldn't locate on slide what this meant. But maybe it might be possible that there is some 4-way multithreading mode also available, besides the standard SMT2.

 

[inf64]

I suspect that Zen4 to Zen3 is what Zen2 to Zen1(+) was. Double the AVX/cache/ L/S badnwidth, increase the thread count (maybe canned for now, who knows), general IPC +15%, shrink to 5nm. There is a Zen3+ in between for APUs on 5nm as a test vehicle as well. Epic clash ahead in 2022!

 

[RetroZombie]

Weird way for represent SMT, 2X and 4X instead of 64/128 and 64/256?
Otherwise the legend would say logical instead of threads?

 

[Richie Rich]

Who will eat cat food now about SMT4? Where is @Thunder 57 with all those strong words? He suddenly disappeared. Perhaps he went to Wallmart to buy some cat food cans....

I told you that SMT4 is on their technology map and it's just a matter of time which core will bring it. And server grade Zen3 with its completely new uarch is god damn good opportunity. And Golden Cove too to be honest.

 

[JoeRambo]

I told you that SMT4 is on their technology map and it's just a matter of time which core will bring it. And server grade Zen3 with its completely new uarch is god damn good opportunity. And Golden Cove too to be honest.

Typical BS aside, for example ProcessLasso ( a great tool to deal with Windows Scheduling with >64 logical cpu systems ) latest changelog:

Process Lasso Revision History

Latest Versions Process Lasso v (changes) Downloads Windows 64-bit Process Lasso (or Beta) Process Lasso Server Edition (or Beta) Windows 32-bit Process Lasso (or Beta) Process Lasso…

bitsum.com

contains this bit:

(11) Core: Support x-way SMT (>2 logical processors per core), for future processors

It might be completely random future looking change, but it might be these guys getting wind from certain vendors about future products or even having such systems in-house.

 

[DrMrLordX]

I told you that SMT4 is on their technology map

You told us it would be in Milan.

 

[Richie Rich]

Typical BS aside, for example ProcessLasso ( a great tool to deal with Windows Scheduling with >64 logical cpu systems ) latest changelog:
It might be completely random future looking change, but it might be these guys getting wind from certain vendors about future products or even having such systems in-house.

I'm the only one who was telling SMT4 might be part of Zen3 since leaked it's new 19h family and it's completelly new uarch (Norrod from AMD). Me and Nosta was telling that Zen1 & Zen2 is just hybrid of Apple's A7 Cyclone and BD as a temporary survival CPU mixture done by Keller (who came from Apple). And that Zen3 is Keller's finally finished Alpha EV8 with SMT4 (Keller is ex-DEC/Alpha engineer) and developed in parallel with Zen2. Maybe you should read entire thread first.

 

[NostaSeronx]

imho, AMD will be killing off SMT rather than increasing it.

=> Kill SMT
=> Switch to double FE + 1C/1T L0i.

Zen3 will launch with "SMT2" but I have been hearing it is actually a VMT2 implementation w/ ST mode being best overall perf/watt.
Zen4 will then drop multithreading on a single core and push for double piped front-end and an improved singlethreaded pure-L0i(no switching between op-cache and L0i).

There might be introductions with something as a Sub-CCX where the L2 is shared between multiple cores. Thus, reducing inter-core latency to the faster L2 from L3 cache.

For example, far down the road =>
16-core/16-thread CCX w/ 64 MB L3 (Local 16 MB blocks of 4 MB per sub-CCX)
4x 4-core/4-thread Sub-CCX w/ 4 MB L2 unified (Local 2 MB blocks of 512 KB and To-CCX 2 MB blocks of 512 KB)
Zen_[blank] core -> 512 KB L2 block -> 512 KB L2 block -> 4 MB block -> 4 MB block -> I/O Fabric

AMD's general trend for the Zen-lineage is the smaller core that improves IPC and EPI is preferred over going bigger >6-wide, SMT4/SMT8, longer SIMDs(512-bit units/256-bit units) which isn't preferred.