Speculation: Ryzen 3000 series

[Topweasel]

Ryzen3 maybe but Navi is earliest end of 2019 probably 2020. Else the Radeon VII wouldn't make that much sense.

I think people were hoping Navi to hit in 2019, early 2019, because they expected 7nm VC option this year and I like others believe that Vega 20 was WS/Server only. But the roadmaps for Navi always had it shaded mostly in the 2020 side of the year. I think people also didn't like the idea of 2021 for the new Arch. Honestly I think Navi gets announced with the PS5/XboxGen4.

 

[trollspotter]

Why not if Navi will be at max a 2060/Vega64 performance-wise? Or if it's just a midrange-GPU like Polaris?

And Lisa said there will be more 7nm GPUs this year, so...

For Ryzen3000 I'm pretty sure we'll see a 16C. And Launch @Computex.

I'm betting the 3700x will be 12c for initial launch. 16c can come later when 7nm yields improve. The top binned perfect dies will be going to server first, and they will probably stockpile the rest for the TR launch late this year.

 

[Spartak]

Q2/Q3 launch gives Intel plenty of time to come up with a tweak to their offerings. Plenty.
The lack of clock speed detail tells me it's not as great as the rumors were saying.

What the hell? They showed a 75W CPU beating the best desktop CPU Intel has for offer that draws almost double the power when you exclude IO. And you find this disappointing?

Performance is EXACTLY as expected hoped for. That gap between 75W and 105-135W TDP will most definately be filled with extra performance by their top of the line products.

 

[DrMrLordX]

Probably June, though April would be great.

Yeah that looks like the window that's been on the table the whole time. I'm still hoping for April but . . . you know, disappointment springs eternal.

Looks like I got the ES clocks about right (~4.6 GHz top) and was right to be skeptical about 16c products on AM4. For now. Nice performance otherwise.

Bring on that Comet Lake Intel! Can't wait to see 10c/20t on 14nm.

 

[Mopetar]

I'm betting the 3700x will be 12c for initial launch. 16c can come later when 7nm yields improve. The top binned perfect dies will be going to server first, and they will probably stockpile the rest for the TR launch late this year.

They might even hold off on launching anything above 8 cores initially. If supplies are tight, they might be able get more value out of using two 6-core dies to sell two CPUs instead of combining them into a single CPU.

It really depends on balancing wafers from TSMC against GF due to the WSA changes. If they're using extra wafers at TSMC for Vega/Navi production, then they probably don't care about having to use extra IO dies.

 

[tamz_msc]

My first guess would be this cinebench is running arond 4.2-4.3 GHz on all cores.

A 9900K at 4.3 GHz doesn't score 2030cb, but does so at 4.7 GHz.

 

[trollspotter]

They might even hold off on launching anything above 8 cores initially. If supplies are tight, they might be able get more value out of using two 6-core dies to sell two CPUs instead of combining them into a single CPU.

It really depends on balancing wafers from TSMC against GF due to the WSA changes. If they're using extra wafers at TSMC for Vega/Navi production, then they probably don't care about having to use extra IO dies.

True, and it really comes down to how compelling the ST performance is. If the ST perf is still lagging Intel by >10% then I suspect we will see 12c+ for launch, but if they are at near parity I could see them delaying anything above 8 cores until late 2019.

 

[Zapetu]

Huh!?

Hasn't that guy been pretty clear throughout that he is basing his reporting on supposed leaks and rumors? I don't recall him stating unequivocally that something absolutely will be the case.

When did he literally lie?

I don't believe Jim lied, but he did get played.

For reference, his source told him there is no other I/O die being produced. It fed him clocks ranges that are now clearly targets at best, and most importantly convinced him there would be a 16C/32T product on the CES show floor today.

He said all a long to take it with a pinch of salt and that it was all rumors, leaks and speculation. He also said in his video that the latest rumors were dropped on him on email and he choose to believe them. His previous speculation about chiplets and IO dies has been very interesting and turned out to be mostly true. These latests rumours about AMD announcing 7nm Ryzen processors (models, clock speeds, prices, etc.) at CES 2019 turned out to be completely wrong. (Edit: to be fair, AdoredTV never said that we would get clock speeds or prices at CES but he led his viewers to believe that AMD would announce a lot more information than they finally did). Also AMD didn't want to give any new information about Navi yet. (Edit: AdoredTV speculated that AMD would announce Navi instead of Vega 2 (Radeon VII) but he did say that it was his own speculation).

I feel a little sorry for Jim (AdoredTV) since his best videos (with interesting technical speculation) haven't got that many views and that's what he does best. He has been wrong in some major points in the past too but this thing got a little out of hand. The most important thing for him to learn about this is not to set expectations too high especially for things like early launches that didn't happen. I still like the guy and I hope this didn't hurt him too much.

We should also give chiakokhua some credit for his early chiplet speculation on AM4 platform which turned out to be partly true (edit: partly because the physical layout is different but the core idea/concept turned out to be true). Sure, AMD didn't show anything with two chiplets yet but atleast chiplet and IO die design is confirmed now.

What I'm most interested about is that chiplet is that is it the exact same chiplet as in Rome. In my opinion it should be but Anandtech's measurements are little different than what we previously estimated by Rome screen shots. The image is very blurry and not well focused so let's investigate a little more later on. If that Ryzen IO die is a little bigger than expected then maybe it holds some Navi chiplet related GPU stuff like display engine and maybe something else. But that's pure speculation and we'll see in the the future what the actual product line will be.

I was very happy that Lisa Su actually showed us the die configuration of the 7nm Ryzen since also that speculation has been very wild for the last couple of weeks. I'm sorry for those who waited for early launch but we'll just have to wait a little longer. The pricing of the Radeon VII (logo with V and two little I's, a little play with roman numerals) was actually cheaper than I expected for what it is. $699 for Vega 20 GPU and 4 stacks of HBM2 memory is not actually that much for what it is. That can only mean good things for price points of the other AMD 7nm products.

 

[Spartak]

How about the 5 GHz 16 core stuff everyone was so in doubt about? Well, the 8c part was likely running somewhere close to that frequency (or at least in the high 4 GHz range), probably with a bit of extra voltage to assure stability in a demo setting, and was only drawing about 75W at full load by AT's rough estimate.

The 2700X scores 1700 at 4.3GHz. If this would be in the upper 4GHz range IPC increase would be zero to negative.

if you take into account the massive improvement in FPU resources, frequency of this chip was 4.3GHz MAX, probably closer to 4GHz.

 

[PotatoWithEarsOnSide]

My first guess would be this cinebench is running arond 4.2-4.3 GHz on all cores.

If we assume the IO consumes around 15W the 8core chiplet runs around 60W or about 7.5W per core.

This _again_ leaves me wondering why they didnt go for 6 core chiplets. 12 cores would give you 90W for the chiplet and 105W total, which is identitical to the current 2700X TDP.
For an extreme edition 125W you could up the all core frequency a bit like she suggested there is room for right now.

If they remain at 8 core tops, and they squeeze another 10% frequency out of it (~4.7GHz), you use that same +40% power budget but instead of 50% extra performance you get 10% extra performance.

OTOH if you put a double chiplet inside this, you either blow your power budget (~135W at these frequencies) or your highest end CPU's need to be significantly clocked down in all core turbo mode. Which on the desktop is quite unusual and would be a expensive waste of die-space. The 3850X as leaked would fit that power envelope with these frequencies though, but leaves all core performance on the table for all but the most exotic systems.

This cinebench demo AGAIN shows 12 core to be the sweet spot for desktop. I cannot see them stay at 8 cores UNLESS this core was clocked significantly low (~4GHz range) and final frequencies will be considerably higher.

The IO will be responsible for more than that; IF was taking up 40-60w on Zen+.
The cores themselves probably run around 4-5w each.

 

[rbk123]

Huh? No, it doesn't.

Huh? Yep, it does.

 

[moinmoin]

I like Jim/AdoredTV as his speculations help me thinking more outside the box regarding the possibilities, like AMD is doing with these designs. But I prefer discussing about the designs and possible ways onward from there, not about what Jim/AdoredTV got right or wrong and whether the latter constitute lies.

Huh? Yep, it does.

I certainly hope they stay competitive.

 

[Tup3x]

So... Basically they showed the 8 core version. They didn't say that it was the highest core count version and looking at the package, it's pretty clear to me that there's space for second chip.

 

[exquisitechar]

Huh? Yep, it does.

Alright, explain what you think Intel can do in that timeframe.

 

[PotatoWithEarsOnSide]

Why not if Navi will be at max a 2060/Vega64 performance-wise? Or if it's just a midrange-GPU like Polaris?

And Lisa said there will be more 7nm GPUs this year, so...

For Ryzen3000 I'm pretty sure we'll see a 16C. And Launch @Computex.

Is there a point to Navi if it is only at 2060 performance when Radeon VII is around 2080 performance?

 

[Spartak]

A 9900K at 4.3 GHz doesn't score 2030cb, but does so at 4.7 GHz.

? We are talking about the Zen2 chip. Both chips werent running at the same clockspeed.

I think you are conflating 'equal setup' with 'clock for clock'.

 

[itsmydamnation]

The 2700X scores 1700 at 4.3GHz. If this would be in the upper 4GHz range IPC increase would be zero to negative.

if you take into account the massive improvement in FPU resources, frequency of this chip was 4.3GHz MAX, probably closer to 4GHz.

CINEBENCH only uses 128bit ops, 256bit units makes no difference.

its either more instructions a clock or more clocks or a bit of both.

Huh? Yep, it does.

intel cant really do anything, they can trade more power for a few more mhz.
even if they bring 10 cores its obvious that amd can do 12 or 16 cores if they wish.

personally i think AMD will only got to 12 cores for no other reason then to go to 16 for Zen3. Otherwise it becomes very hard to get a good perf uplift for the gen after Zen2 thats still on some form of 7nm (EUV).

 

[Fanatical Meat]

So any reasonably firm release date or is it just at 2nd half 2019?

 

[epsilon84]

Came here expecting a 16C Ryzen 3000 announcement so consider me a bit underwhelmed...

Btw, Zen+ already beats CFL clock for clock in CB15. A 4.5GHz 2700X would probably be enough to match a 9900K, FWIW.

Assuming a 10% IPC uplift from Zen2, based on the 2050 CB15 score, I'd estimate that the CES sample isn't running much higher than 4GHz.

 

[tamz_msc]

? We are talking about the Zen2 chip. Both chips werent running at the same clockspeed.

I think you are conflating 'equal setup' with 'clock for clock'.

I don't think we are in a position to guess what clock speeds the Zen 2 sample was running at.

 

[PotatoWithEarsOnSide]

I think that the main takeaway from the CB demo was that an 8c Ryzen 3xxx matched an 8c 9900K at ~4.7GHz. The implication is either the IPC is ahead, the clocks are ahead, or both are ahead but to a lesser extent.
The 9900K clearly wasn't gimped; it's CB score was comparable to those posted publicly by owners in the past.

 

[DrMrLordX]

So any reasonably firm release date or is it just at 2nd half 2019?

More like late Q2/early Q3, probably June. 2H implies as late as December which is too late for Zen2 based on the mid-year release.

For those talking about AdoredTV, I don't see much fundamentally WRONG with what he predicted. The room is there on the PCB for an additional chiplet. That tells us that 16c AM4 chips are possible. AMD just didn't demo one. They also said nothing about clocks. It also tells us that that GPU and GPU+HBM blocks are possible for APUs to come later. The I/O die gives AMD a lot of flexibility.

 

[Spartak]

Looks like I got the ES clocks about right (~4.6 GHz top) and was right to be skeptical about 16c products on AM4. For now. Nice performance otherwise.

Bring on that Comet Lake Intel! Can't wait to see 10c/20t on 14nm.

Clocks werent mentioned, just that the 9900k ran at stock.

if this CPU with double the FP resources accross the board only manages to extract ~10% IPC gain in FP operations, that would be a rather big disappointment on the architecture side.

 

[DrMrLordX]

Clocks werent mentioned, just that the 9900k ran at stock.

if this CPU with double the FP resources accross the board only manages to extract ~10% IPC gain in FP operations, that would be a rather big disappointment on the architecture side.

Double FP performance across the board only applies to AVX2. Which CBR15 does not use.

My general assumption Is that the ES is running in the 4.0-4.3 GHz range, so we should not expect more than 300-400 more MHz out of a launch Zen2 product for top boost clocks. Feel free to prove otherwise, AMD!

 

[itsmydamnation]

Clocks werent mentioned, just that the 9900k ran at stock.

if this CPU with double the FP resources accross the board only manages to extract ~10% IPC gain in FP operations, that would be a rather big disappointment on the architecture side.

know your workload......