Speculation: Ryzen 3000 series

Page 48 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

What will Ryzen 3000 for AM4 look like?


  • Total voters
    230
Dec 10, 2018
63
84
51
To me, and this is just my opinion, the simple fact that the CPUs are H2 2019 makes the whole table false just by the fact it has way too much information too soon. And the 240GE at $75 kinda makes the $99 6C/12T very very unlikely.

But still it may end up correct in a few cases, especialy SKU names and core configuration.

Is there any technical reason why 6C/12T is unlikely at $99? The 240GE isn't on TSMC 7nm so it's hard to just compare the two like that. If Zen2 was still fabbed at GF then sure, I think it'd be fair to call bs and say 6C/12T won't be 99$ because the 240GE is 75$. But, it's a new process with smaller die area, and probably relatively high yielding.
 

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
Is there any technical reason why 6C/12T is unlikely at $99? The 240GE isn't on TSMC 7nm so it's hard to just compare the two like that. If Zen2 was still fabbed at GF then sure, I think it'd be fair to call bs and say 6C/12T won't be 99$ because the 240GE is 75$. But, it's a new process with smaller die area, and probably relatively high yielding.

No technical reasons, they may do it is they want. The question here is "AMD needs to launch a 6C/12T at $99 coming from a 4/4 completely skipping 4/8 and 6/6 at that price range?" I think not.

The 240GE was important for the table because it was launched at late december after the table went live, so it was very hard to belive that AMD would launch a $75 dual core in December and then 1-3 months later a $99 6C/12T.

And remember they need to set up a entire lineup that makes sence, below $100 there are the Athlons and right now the top 2C/4T Athlon is $75, if they are going to sell the $99 6C/12T CPU, were the 4/4 and 4/8 APUs will fit in there? They would launch then as sub $100 Athlons when there is no other APU in the near future? I dont think so, they are going to lose a lot of money if they do that. They are cutting down their profits significaly there.

Now if you tell me that the 6C/12T will take the 2400G place at $140, the 4/8 APU the 2200G place at $100 and the 4/4 APU the 240GE place at $75 this make a lot more sence to me.
 
  • Like
Reactions: CHADBOGA
Dec 10, 2018
63
84
51
No technical reasons, they may do it is they want. The question here is "AMD needs to launch a 6C/12T at $99 coming from a 4/4 completely skipping 4/8 and 6/6 at that price range?" I think not.

Well of course AMD doesn't need to do that. However the argument for why it'd be in their interest is to grab more market share from Intel. AMD probably won't be losing much from pricing their chip lower, and if the chips sell well enough, they might even earn more than if they kept pricing the same. AMD pricing for zen1 was amazing already compared to Intel so it's not bad speculation to say they can and will do it again.

The 240GE was important for the table because it was launched at late december after the table went live, so it was very hard to belive that AMD would launch a $75 dual core in December and then 1-3 months later a $99 6C/12T.
That's true; it would seem a bit strange. I also wouldn't be surprised if AMD is trying to sell the Athlon to OEMs to be used in their cheapest systems.

And remember they need to set up a entire lineup that makes sence, below $100 there are the Athlons and right now the top 2C/4T Athlon is $75, if they are going to sell the $99 6C/12T CPU, were the 4/4 and 4/8 APUs will fit in there? They would launch then as sub $100 Athlons when there is no other APU in the near future? I dont think so, they are going to lose a lot of money if they do that. They are cutting down their profits significaly there.

Now if you tell me that the 6C/12T will take the 2400G place at $140, the 4/8 APU the 2200G place at $100 and the 4/4 APU the 240GE place at $75 this make a lot more sence to me.

I don't think the same argument holds here because APUs are a different market segment. Most in the market for an APU probably won't buy Ryzen 3k (because they don't want to spend more on a dGPU), and I don't think AMD was ever trying to convince people looking for a CPU to buy an APU. Also, the 2k APUs are starting to fade away as "last gen" and I don't think AMD would cry much from lower sales there. They're going to be replaced by 3k APUs in the next year, and I almost expect AMD to pull of something new and unexpected (heh) with their chiplet strategy for that.
 

Mockingbird

Senior member
Feb 12, 2017
733
741
106
But AMD don't want them to make up a big chunk of their revenue.

Yes they want to keep selling to gamers - but they'd much rather a massive chunk of their revenue was coming from server/workstation markets - as that is where the real profits are.

I am talking about desktop processors only.

Server processors (EPYC) are obviously going to be priced differently.

I literally was talking about development costs in my first point.

Gamers may make up a significant fraction of AMDs marketshare at the moment, but like i said and which has been mentioned earlier in the thread, there are much larger market segments that AMD is in a prime position to gain a share in; AMD realizes this as well and they'd be fools not to take advantage of it. Intel is having supply issues with 14nm, a stagnant architecture, and higher price.

How do they further incentivize that market to buy their cpus instead of intels?

Offer 50% more cores at the same or lower price, at comparable or greater performance, and with greater power efficiency (as we saw with the ES demo).

Even if AMD has lower margins with those prices, it's probably a trade-off they're willing to take to gain market share. They can improve their margin by improving yield or in future products.

I'm not sure that incorrect pricing necessarily invalidates the entirety of the leak.
I'm also not convinced that their existing best seller would continue to be their best seller, though I can see your logic in that it could potentially lower the ASP which might in turn "hurt" AMD. However, reduced margins on increased sales can still produce significantly more profit, both short and long term, so even that "hurt" is relative; reduced margins on increased volume actually hurts Intel more than it would AMD, as they are the ones losing sales and margins.
In relation to the second part of your post, I can see this happening if the performance gain from one generation to the next is above and beyond what we've become used to. However, I suspect that your R9 pricing may be a little on the high side; the goal won't be to maximise margins, rather to decimate Intel's own high margin sales, and this is best done by beating Intel on all three metrics by a distance (price, performance, and performance per watt).

Well of course AMD doesn't need to do that. However the argument for why it'd be in their interest is to grab more market share from Intel. AMD probably won't be losing much from pricing their chip lower, and if the chips sell well enough, they might even earn more than if they kept pricing the same. AMD pricing for zen1 was amazing already compared to Intel so it's not bad speculation to say they can and will do it again.


That's true; it would seem a bit strange. I also wouldn't be surprised if AMD is trying to sell the Athlon to OEMs to be used in their cheapest systems.



I don't think the same argument holds here because APUs are a different market segment. Most in the market for an APU probably won't buy Ryzen 3k (because they don't want to spend more on a dGPU), and I don't think AMD was ever trying to convince people looking for a CPU to buy an APU. Also, the 2k APUs are starting to fade away as "last gen" and I don't think AMD would cry much from lower sales there. They're going to be replaced by 3k APUs in the next year, and I almost expect AMD to pull of something new and unexpected (heh) with their chiplet strategy for that.

AMD already dominate the ~$200 and under (processor) market.

It doesn't make sense for AMD to drop the prices further to compete with itself.

It's the $300+ (processor) market that Intel dominate with products such as the Core, i7-8700K, Core i7-9700K, Core i9-9900K
 
Last edited:

Mockingbird

Senior member
Feb 12, 2017
733
741
106
Is there any technical reason why 6C/12T is unlikely at $99? The 240GE isn't on TSMC 7nm so it's hard to just compare the two like that. If Zen2 was still fabbed at GF then sure, I think it'd be fair to call bs and say 6C/12T won't be 99$ because the 240GE is 75$. But, it's a new process with smaller die area, and probably relatively high yielding.

What you got wrong is that you think that you can just take something, and die shrink it to make it cheaper.

That used to be universally true back in the day, but the cost of smaller processes are becoming so expensive that is no longer necessary true.

amd-iedm-2017-23-1024x579-png.134744
 

Spartak

Senior member
Jul 4, 2015
353
266
136
That's my point; he was right about the chiplets!

You have any technical insight/arguments against his "leaks" (whatever that means to you)? I'd love to hear them because I don't know enough about TSMC's 7nm process, and AMD's metrics on power/frequency/feature scaling to argue for or against them myself. I'm just here to talk and learn more about the architecture.

And again his chiplet prediction had nothing to do with that "leaked" list. If you cannot distinguish between both I don't know what to say. There's no point.

And regarding the technical arguments, I already mentioned them.
(1)They werent announced at CES, (2)the TDP for R9 is wrong and (3)there will be no G-versions until Renoir, both of the latter two are confirmed by AMD. These are all known FACTS.

The only remaining two elements are frequencies and pricing. For final frequencies its way to early, this was even mentioned on stage before the demo (still not final frequencies).

That leaves pricing, which seems very unlikely on the bottom end. The top of the line is just informed speculation that I think everyone would expect based on the Zen1 R7 price points.

It's just a fantasy list people, let it go.
 
Last edited:

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
What people are asking you for is technical reasons why XYZ aren't possible, as that's all anyone is really interested in.
No-one cares whether the detail is spot on, only whether the details as presented are plausible.
If you'd followed the discussion, potential core counts, TDPs, and pricing has already been discussed in depth, and nothing suggests that the information provided was categorically wrong.
The only thing we don't really know anything about is the likely clocks on the various SKUs, since there's no common ground on the clocks for the ES, or how much higher a production CPU may or may not have.
Sure, nothing concrete was announced at CES regarding specific SKUs, but that's the least interesting of all the speculation in this thread.

There's very little doubt that a) Ryzen could end up with a 16c SKU, b) TDPs can be within the speculated ranges quite comfortably, and c) chiplet design will have a large impact on reducing per unit production costs, hence the possibility of low prices. Yes, we've not had 100% confirmation of any of the above, and we're not expecting to get that any time soon, but at least we know that it is plausible. That's all we care about at this stage.
 

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
What you got wrong is that you think that you can just take something, and die shrink it to make it cheaper.

That used to be universally true back in the day, but the cost of smaller processes are becoming so expensive that is no longer necessary true.

That's the brilliance of AMD's approach though. Using chiplets, they can achieve much higher yields than they would otherwise be able to with a monolithic die.

The cost factor doesn't just increase as a result of the newer process being more expensive (on account of needing to recoup investment in its development) but also because it's immature. If you make a 250 mm^2 die on a new process, you might get fewer than a third of your chips back without defects. You can probably salvage a lot of the defective dies regardless, but they can't be sold for as much as a full die.

However, switching to the chiplet-based approach that AMD uses on that same process means that they can get closer to 70% full dies from the same wafer simply because the chiplets are much smaller (~80 mm^2) and less likely to have a defect. We're not very concerned with the IO dies since they're on the old, mature process so yields should be good.

This completely changes the economics and makes it feasible for AMD to continue to lower costs, even though the new node wouldn't normally allow that if it were just a matter of shrinking everything down.
 

Spartak

Senior member
Jul 4, 2015
353
266
136
What people are asking you for is technical reasons why XYZ aren't possible, as that's all anyone is really interested in.
No-one cares whether the detail is spot on, only whether the details as presented are plausible.
If you'd followed the discussion, potential core counts, TDPs, and pricing has already been discussed in depth, and nothing suggests that the information provided was categorically wrong.
The only thing we don't really know anything about is the likely clocks on the various SKUs, since there's no common ground on the clocks for the ES, or how much higher a production CPU may or may not have.
Sure, nothing concrete was announced at CES regarding specific SKUs, but that's the least interesting of all the speculation in this thread.

There's very little doubt that a) Ryzen could end up with a 16c SKU, b) TDPs can be within the speculated ranges quite comfortably, and c) chiplet design will have a large impact on reducing per unit production costs, hence the possibility of low prices. Yes, we've not had 100% confirmation of any of the above, and we're not expecting to get that any time soon, but at least we know that it is plausible. That's all we care about at this stage.

True all that. But we didnt need that list to come to that conclusion ourselves. Many people in the start of this very thread predicted that before AdoredTV multi-die speculation got conflated with the 'leaked' list. There's still people like somethingclever that cannot distinguish between both.

People were psyched up about that "leak" because (1) it seemed to launch earlier then expected (2) frequencies were much higher then expected (3) prices much lower at the low-end

1. has been proven wrong (launch = mid '19 the soonest)
2. still completely up in the air
3. highly unlikely at the bottom-end.
 
Last edited:

Mockingbird

Senior member
Feb 12, 2017
733
741
106
That's the brilliance of AMD's approach though. Using chiplets, they can achieve much higher yields than they would otherwise be able to with a monolithic die.

The cost factor doesn't just increase as a result of the newer process being more expensive (on account of needing to recoup investment in its development) but also because it's immature. If you make a 250 mm^2 die on a new process, you might get fewer than a third of your chips back without defects. You can probably salvage a lot of the defective dies regardless, but they can't be sold for as much as a full die.

However, switching to the chiplet-based approach that AMD uses on that same process means that they can get closer to 70% full dies from the same wafer simply because the chiplets are much smaller (~80 mm^2) and less likely to have a defect. We're not very concerned with the IO dies since they're on the old, mature process so yields should be good.

This completely changes the economics and makes it feasible for AMD to continue to lower costs, even though the new node wouldn't normally allow that if it were just a matter of shrinking everything down.

The point of going with the chiplet design is to mitigate the price increase associated with 7nm.

Yes, you managed to mitigate the price increase, but it's still a price increase.
 

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
Whilst that is true, the R&D overheads will be spread out over a much larger volume if sales. Ultimately then it comes down to whether the savings per unit of one will offset the other. I suspect that they would.
That $400m R&D cost, versus $150m last time, is likely to be split over way more than 3x the volume of sales, especially at the higher margin end of the table...which ch ultimately is where all the gravy is at.
 
  • Like
Reactions: somethingclever

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
To me, and this is just my opinion, the simple fact that the CPUs are H2 2019

Mid-2019, probably June (May if we are lucky; don't hold your breath). That's H1.

And the 240GE at $75 kinda makes the $99 6C/12T very very unlikely.

Why? The 240GE (and all the other GE Raven Ridge products) are low-volume, low-cost chips that only serve a tiny niche anyway. People that won't know/won't care when AMD launches Ryzen 3 CPUs. Do you really think AMD is that worried about the 240GE that they'll go out of their way to avoid orphaning whatever stock is left in the channel?
 
  • Like
Reactions: trollspotter

Mockingbird

Senior member
Feb 12, 2017
733
741
106
The industry is at a crossroads.

It used to be that you can take something, die shink it, and now it's cheaper.

You can then go back to the drawing board and come up with a more complex design.

We have never before had a time when this wasn't true.
 
  • Like
Reactions: PeterScott

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
Why? The 240GE (and all the other GE Raven Ridge products) are low-volume, low-cost chips that only serve a tiny niche anyway. People that won't know/won't care when AMD launches Ryzen 3 CPUs. Do you really think AMD is that worried about the 240GE that they'll go out of their way to avoid orphaning whatever stock is left in the channel?

Because the 240GE at $75 and 3300 6C/12T at $99 makes it impossible to fit the quad core APUs in the lineup.

Quad core APUs would be all sub$100 meaning cutting down the 2400G price by half when there is no other APUs in the horizon, is too much. AMD may not care about sub$100 Athlons but you can bet they care about the 2200 and 2400G. With no new APUs dont expect RR prices to change too much, the 2400G dropping to $100 would be impresive already, and that alone may be too much.

All of this only makes sence if there were new Zen 2 APUs and thats is not happening any time soon...
Unless... All Zen 2 AM4 are APUs and the I/O die is hidding a Vega 11 inside, what would be a bit difficult.
 
Last edited:
  • Like
Reactions: Spartak

DarthKyrie

Golden Member
Jul 11, 2016
1,531
1,279
146
Because the 240GE at $75 and 3300 6C/12T at $99 makes it impossible to fit the quad core APUs in the lineup.

Quad core APUs would be all sub$100 meaning cutting down the 2400G price by half when there is no other APUs in the horizon, is too much. AMD may not care about sub$100 Athlons but you can bet they care about the 2200 and 2400G. With no new APUs dont expect RR prices to change too much, the 2400G dropping to $100 would be impresive already, and that alone may be too much.

All of this only makes sence if there were new Zen 2 APUs and thats is not happening any time soon...
Unless... All Zen 2 AM4 are APUs and the I/O die is hidding a Vega 11 inside, what would be a bit difficult.

You do know that that $75 chip is EOL and will be replaced by a 3000 series chip at what will probably be a lower price point. Do you honestly think that AMD is going to continue to produce the 2000 series APUs on 14nm when the 3000 series APUs are being produced on 12nm? I don't know what to tell you if you actually believe this.
 
Dec 10, 2018
63
84
51
What you got wrong is that you think that you can just take something, and die shrink it to make it cheaper.

You're confusing process node with die size, and manufacturing costs with development costs.

A smaller die will always decrease manufacturing costs because of higher yield, more chips per wafer and overhead being spread over a greater volume. Cost per chip goes down.

A smaller process increases developmental cost. What AMD did to minimize that is having only 1 7nm die and the chiplet strategy.
 
Dec 10, 2018
63
84
51
And again his chiplet prediction had nothing to do with that "leaked" list. If you cannot distinguish between both I don't know what to say. There's no point.

Like I said we don't care... The discussion here is speculation. That leaked list is speculation and we can talk about frequencies and TDPs while not caring about it they're correct or not too.

And regarding the technical arguments, I already mentioned them.
(1)They werent announced at CES, (2)the TDP for R9 is wrong and (3)there will be no G-versions until Renoir, both of the latter two are confirmed by AMD. These are all known FACTS

Those are not technical reasons. Why is the TDP wrong? What about TSMC 7nm or the zen2 architecture makes that TDP so unlikely?

Sure no G versions until later. No one's disagreeing with you there.

That leaves pricing, which seems very unlikely on the bottom end. The top of the line is just informed speculation that I think everyone would expect based on the Zen1 R7 price points

Yeah we were having a nice discussion about that! You have anything to contribute?
 
Last edited:

Mockingbird

Senior member
Feb 12, 2017
733
741
106
You do know that that $75 chip is EOL and will be replaced by a 3000 series chip at what will probably be a lower price point. Do you honestly think that AMD is going to continue to produce the 2000 series APUs on 14nm when the 3000 series APUs are being produced on 12nm? I don't know what to tell you if you actually believe this.

It gets replaced by the desktop version of the Ryzen 3 3300U, which is basically the same thing except with higher clock speeds, at the same price.
 
Last edited:
  • Like
Reactions: PeterScott

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
You do know that that $75 chip is EOL and will be replaced by a 3000 series chip at what will probably be a lower price point. Do you honestly think that AMD is going to continue to produce the 2000 series APUs on 14nm when the 3000 series APUs are being produced on 12nm? I don't know what to tell you if you actually believe this.

EOL? It launched like 3 weeks ago, the 240GE will be reeplaced, sometime after the Ryzen 3000 launch, i would not expect the Athlons 300 before Q4 2019. It took 6 months launch the Athlon 200GE after the 2400G.
 
Last edited:

Mockingbird

Senior member
Feb 12, 2017
733
741
106
What I don’t get is this:

Over half a year before the products will be announced, a dubious list appears and people, for whatever reason, treat it like it has some legitimacy.

Never mind that AMD don’t usually prices and names its products until close to release (usually 1-2 weeks prior).

Never mind that none of the products that were supposedly going to be announced at CES were announced.
 
Last edited:

trollspotter

Member
Jan 4, 2011
28
35
91
EOL? It launched like 3 weeks ago, the 240GE will be reeplaced, sometime after the Ryzen 3000 launch, i would not expect the Athlons 300 before Q4 2019. It took 6 months launch the Athlon 200GE after the 2400G.
The 240GE will have been on the market for about 6-7 months by the time Ryzen 3000 launches, and will still sell for a few more months beyond that, probably with a steep price cut. It is just a way for them to sell a few more chips that otherwise would have been thrown away. It will have been on the market for around 8-10 months, which is plenty for a salvaged part that doesn't make them much money, and for which production should soon be ending.
 

Mockingbird

Senior member
Feb 12, 2017
733
741
106
The 240GE will have been on the market for about 6-7 months by the time Ryzen 3000 launches, and will still sell for a few more months beyond that, probably with a steep price cut. It is just a way for them to sell a few more chips that otherwise would have been thrown away. It will have been on the market for around 8-10 months, which is plenty for a salvaged part that doesn't make them much money, and for which production should soon be ending.

2nd generation Ryzen desktop APUs just get replaced by 3rd generation desktop Ryzen APUs based on the 3rd generation Ryzen Mobile processors that AMD announced over a week ago.

Athlon 340GE would be pretty much the same as Athlon 240GE, except with a slightly higher clock speed.

*note: Athlon 240GE isn't branded Ryzen, but is considered a Ryzen APU in this content.
 
  • Like
Reactions: PeterScott

Mockingbird

Senior member
Feb 12, 2017
733
741
106
Expectation:

Ryzen 3 3200G - 4C/4T - $99

Ryzen 3 3400G - 4C/8T - $149

Ryzen 5 3600 - 6C/12T - $199

Ryzen 5 3600X - 6C/12T - $229

Ryzen 7 3700 - 8C/16T - $299

Ryzen 7 3700X - 8C/16T - $329

Ryzen 9 3800 - 12C/24T - $499

Ryzen 9 3800X - 12C/24T - $529

Ryzen 9 3900 - 16C/32T - $699

Ryzen 9 3900X - 16C/32T - $729