Speculation: Ryzen 3000 series

[Mockingbird]

I really can't see how you keep stating this as fact. When you say AMD dominates the $200 and lower market, what do you mean? Do you really believe that AMD sells more CPUs than intel in that range? Seeing that most CPU sales are ones less that $200, then are they already outselling Intel?

Yes.

Let's look at Intel's processors under $200.

There's the Core i5-8400, Core i3-8350K, Core i3-8100, and the Pentium processors.

Ryzen 5 2600 is the best selling processor at pretty much every retailers and easily our sell the Core i5-8400

Ryzen 3 2200G outsells the Core i3-8100.

Core i3-8350K is a horrible seller.

Maybe the Pentium outsell the Athlon that just came out, but that's it.

Note that I am talking about retail processor sales, not OEM sales.

 

[somethingclever]

I do wonder if AMD would be afraid to put a Vega 3 IGP on the IO die for fear of killing off APU sales. I do think it'd be worth it; since they won''t be getting much Corporate desktop OEM sales without it.

It won't matter as much if ryzen is priced so low that even paired with a cheap dgpu, it still beats Intel in price.

If you disable the 4 slowest cores and only enable the fastest 4 cores, the possibility exists for some ultra fast CPUs relative to a 1 chiplet 8C. A 50th anniversary special edition?

Wouldn't this be difficult/a very small bin? I recall the cores in each ccx were "slaved" (for lack of a better term) to each other.

E.g if you disable core 0 on a chiplet you'd also have to disable core 4 and the same cores on any other chiplets on the same package.

That's why we don't have odd core numbers, and why there's no 14 core chip.

 

[Mockingbird]

I don't get people on this forum who said, "AMD needs disruptive pricing".

If AMD's $229 6C/12 Ryzen 5 3600X can match Intel's $369 Core i7-8700K and AMD's $329 8C/16T Ryzen 7 3700X can match Intel's $529 Core-9900K, that's not "disruptive pricing"?

 

[scannall]

I don't get people on this forum who said, "AMD needs disruptive pricing".

If AMD's $229 6C/12 Ryzen 5 3600X can dethrone Intel's $369 Core i7-8700K and AMD's $329 8C/16T Ryzen 7 3700X can dethrone Intel's $629 Core-9900K, that's not "disruptive pricing"?

Yes, yes it is. Intel has been fighting hard for CPU's to NOT be a commodity. But they are. The secret of CPU's has been out there for a long time. At this particular time, AMD is making better choices. Are there advantages to owning your own fabs? Oh yeah, more than a couple. But there are also advantages to using a fab like TSMC, where a lot of people chip in on node advancement. It's just nice to see actual advancement, instead of milking.

 

[Shivansps]

There is a some talk going on about some leak on the future console APU.

 

[maddie]

It won't matter as much if ryzen is priced so low that even paired with a cheap dgpu, it still beats Intel in price.



Wouldn't this be difficult/a very small bin? I recall the cores in each ccx were "slaved" (for lack of a better term) to each other.

E.g if you disable core 0 on a chiplet you'd also have to disable core 4 and the same cores on any other chiplets on the same package.

That's why we don't have odd core numbers, and why there's no 14 core chip.

Yes, a tiny bin in the grand scheme. I should have made this clearer in the original post on this idea. My fault.

With regards to the disabling options, we have no idea as to the chiplet core layout. What was impossible in Zen1 might not still apply to Zen2. Time will tell.

 

[maddie]

There is a some talk going on about some leak on the future console APU.

Now this evoked a laugh. Please don't be insulted as humor is good.

Some talk on some leak on some console. Got it.

 

[OTG]

If AMD's $229 6C/12 Ryzen 5 3600X can match Intel's $369 Core i7-8700K and AMD's $329 8C/16T Ryzen 7 3700X can match Intel's $629 Core-9900K, that's not "disruptive pricing"?

Copy-pasting the current price/core isn't particularly disruptive, no.
It would still be a great value, make AMD plenty of money, even gain back more market share. But it's just what you'd expect from a new generation on a new node.

Matching Intel's Top Gaming CPU Ever with a $180 R5, then blowing it away with a $329 12-core R7?
While Intel is still trying to sell the 6-core 8700k for $380?
And all along, everybody knows they're hiding a 16-core elephant behind the drapes.
Disruptive.
Will it happen? Obviously I think so, and I think it's AMD's best shot at setting the company up for long-term success.

 

[raghu78]

I don't get people on this forum who said, "AMD needs disruptive pricing".

If AMD's $229 6C/12 Ryzen 5 3600X can match Intel's $369 Core i7-8700K and AMD's $329 8C/16T Ryzen 7 3700X can match Intel's $629 Core-9900K, that's not "disruptive pricing"?

Intel has a 10C/20T Comet Lake likely coming in late Q3 2019. So AMD has to project where the competition will be in their response to Ryzen. btw AMD Ryzen 3k is mostly launching in late Q2 or early Q3 2019. So just around 3 months before the rumoured next gen Comet Lake. AMD has to price their products such that Intel cannot compete with CFL-R and need Comet Lake to even get back at competing.

So I am playing devil's advocate here and guessing how AMD could price and how Intel can respond.

Ryzen 3k
Ryzen 9 - 16C/32T - $499 - $599
Ryzen 7 - 12C/24T - $349 - $399
Ryzen 5 - 8C/16T - $249 - $279
Ryzen 3 - 6C/12T - $149 - $169

Ryzen 2k
Ryzen 7 - 8C/16T - $199 - $219
Ryzen 5 - 6C/12T - $99 - $119

You have to remember that AMD's 7nm supply in 2019 will be limited. New product generation ramps take atleast 2Q to see high volume. So Ryzen 2k will be the much larger portion of 2019 AMD desktop CPU sales. By H1 2020 Ryzen 3k will take over as the larger volume generation.

Intel Comet Lake

10C/20T - $449
10C/10T - $369
8C/16T - $299
8C/8T - $249
6C/12T - $199
6C/6T - $169

The problem for Intel will be if AMD hits max turbo clocks of 4.6-4.7 Ghz with 10% higher IPC than Zen+ which is very much possible than Intel will have no huge ST perf lead and find itself totally outclassed on MT perf. Until Intel can bring 10nm gens with > 10 cores AMD will be in a leadership position on the desktop.

 

[Mockingbird]

Copy-pasting the current price/core isn't particularly disruptive, no.
It would still be a great value, make AMD plenty of money, even gain back more market share. But it's just what you'd expect from a new generation on a new node.

Matching Intel's Top Gaming CPU Ever with a $180 R5, then blowing it away with a $329 12-core R7?
While Intel is still trying to sell the 6-core 8700k for $380?
And all along, everybody knows they're hiding a 16-core elephant behind the drapes.
Disruptive.
Will it happen? Obviously I think so, and I think it's AMD's best shot at setting the company up for long-term success.

@PeterScott said it best:

I am a realist. You are a dreamer.

 

[Mockingbird]

Intel has a 10C/20T Comet Lake likely coming in late Q3 2019. So AMD has to project where the competition will be in their response to Ryzen. btw AMD Ryzen 3k is mostly launching in late Q2 or early Q3 2019. So just around 3 months before the rumoured next gen Comet Lake. AMD has to price their products such that Intel cannot compete with CFL-R and need Comet Lake to even get back at competing.

So I am playing devil's advocate here and guessing how AMD could price and how Intel can respond.

Ryzen 3k
Ryzen 9 - 16C/32T - $499 - $599
Ryzen 7 - 12C/24T - $349 - $399
Ryzen 5 - 8C/16T - $249 - $279
Ryzen 3 - 6C/12T - $149 - $169

Ryzen 2k
Ryzen 7 - 8C/16T - $199 - $219
Ryzen 5 - 6C/12T - $99 - $119

You have to remember that AMD's 7nm supply in 2019 will be limited. New product generation ramps take atleast 2Q to see high volume. So Ryzen 2k will be the much larger portion of 2019 AMD desktop CPU sales. By H1 2020 Ryzen 3k will take over as the larger volume generation.

Intel Comet Lake

10C/20T - $449
10C/10T - $369
8C/16T - $299
8C/8T - $249
6C/12T - $199
6C/6T - $169

The problem for Intel will be if AMD hits max turbo clocks of 4.6-4.7 Ghz with 10% higher IPC than Zen+ which is very much possible than Intel will have no huge ST perf lead and find itself totally outclassed on MT perf. Until Intel can bring 10nm gens with > 10 cores AMD will be in a leadership position on the desktop.

Those prices are way to too generous.

Also, if I were AMD, I would wait until Intel releases the 10C Comet Lake, then release the 12C and 16C Ryzen.

 

[raghu78]

Those prices are way to too generous.

Also, if I were AMD, I would wait until Intel releases the 10C Comet Lake, then release the 12C and 16C Ryzen.

Why would you wait for Intel to launch ? You have to preemptively target the competition's next gen when its just 3 months away. When you are leading the competition on a leading edge node transition you are not expected to be reactive but proactive. Luckily you are not making decisions at AMD. AMD is going to sell their products at good margins while providing much better MT perf/$ . Can Intel compete with a 10C/20T CometLake against a 16C/32T Ryzen 9 3800x. Not at all. Can they do against a 12C/24T Ryzen 7 ? Much better than in the former case. Intel will extract another 200 - 300Mhz on max turbo and try to get to > 4.5 Ghz all core turbo. imo AMD will come up with a Ryzen 9 at the top of the stack to go up against the core i9. Its so blindingly obvious. AMD will want to unequivocally take the desktop perf crown in 2019 and keep it for a few years.

 

[OTG]

@PeterScott said it best:

I am a realist. You are a dreamer.

That's cute, but your 'grasp' of the situation ("intel will just sell their chips for nothing, AMD is doomed" "a price war isn't a price war until somebody starts illegally dumping product") seems to show otherwise.

I'm not a dreamer because I think AMD needs to be super aggressive with pricing, I'm looking ahead to Intel's inevitable return to having a competitive product.
Maybe that's even more realistic than thinking that AMD should rest on their laurels after having a whopping 2 generations of decent CPUs.

 

[beginner99]

You can talk about economic theory all you want, but I think Intel and AMD will be price fixing. If you accept this you will be less affected by this summer's sticker shock.

Agree. Not really price fixing but intel is supply constrained so they are bound to be expensive and hence AMD can price accordingly just like they do with GPUs. 590 and radeon vii aren't exactly great deals as they are simply priced like nv pricing. They would need market share even more in gpu space yet there seems no incentive to do so. they could have offered the 590 for $200. But they did not.

If, on the other hand, AMD goes HAM and prices their 12-core part at $329, their 8-core part at $199, and 6-cores starting at $99, they completely eliminate Intel as a viable option!

See above. it's clear that this is not AMDs strategy because we can see it in the market and because they said so themselves (not wanting to be the cheap brand).

Plus this strategy destroys their own market. Once the prices are that low, people simply won't pay more (unless it's a near monopoly, see NV pricing) and simply won't buy at all until AMD/intel offer better options.

 

[DrMrLordX]

You only have to look at GPU pricing to see AMD chases margins over market share, once it actually has enough market share to have a vested interest in protecting it's margins.

RTG is facing a completely different demand level for their products. Their pro cards are flying off the shelves. The AI market is growing at a staggering pace. It's so extreme that it's difficult for RTG to justify R&D or sales directed to the consumer market of any product that has potential in the pro market. If demand is near-infinite, why not charge insane prices? Why sell anything for less than $1k? The only products AMD really wants to sell to consumers are those with no pro applications: Polaris and (presumably) Navi.

It's killing the "halo" consumer dGPU segment.

No matter well AMD does with Zen2, demand for that product will not reach what they see today for Vega20.

I don't get people on this forum who said, "AMD needs disruptive pricing".

AMD's pricing levels are already distruptive. The 2700x was cheap vs. the 8700k and the 1800x was cheap against the 6800 and 7700k. If they stick to the same script, everything in their Zen2 lineup will look great against Intel's chips. They do not have to change much about their pricing strategy.

 

[PeterScott]

RTG is facing a completely different demand level for their products. Their pro cards are flying off the shelves. The AI market is growing at a staggering pace. It's so extreme that it's difficult for RTG to justify R&D or sales directed to the consumer market of any product that has potential in the pro market. If demand is near-infinite, why not charge insane prices? Why sell anything for less than $1k? The only products AMD really wants to sell to consumers are those with no pro applications: Polaris and (presumably) Navi.

Where is the evidence that AMD Pro GPU cards are "flying off the shelves"?

The real "pro" market is in data centers, which NVidia utterly dominates. It also looks like CUDA dominates GPGPU programming in nearly every application space. OpenCL seems more about hobbyists than the actual Pro market.

Despite NVidia dominating success with pro data center cards, it still makes most of it's money on the gaming cards. For GPUs, the Consumer market is the most important market.

So any claims about lacking justification for consumer R&D fall very flat. Vega looks a lot more like consumer cards trying to break into Datacenter than the other way around.

 

[PeterScott]

Copy-pasting the current price/core isn't particularly disruptive, no.
It would still be a great value, make AMD plenty of money, even gain back more market share. But it's just what you'd expect from a new generation on a new node.

Who is AMD looking to disrupt? Themselves?

If AMD have a comparable product to Intel, and AMD already sell it for significantly less than Intel, then what is gained by lowering it more? Why aren't they already dominating the CPU commodity market with superior price/performance being recognized?

Because it's more likely that AMD has a perception/marketing/education issue, not a pricing issue.

Simply dropping pricing to deal with perception issue, will leave barrels of money the table, and reinforce the perception of AMD as the cheap option, not the premium option.

AMD 6 and 8 core parts, will not decrease in price, when they introduce an even better Zen core, because Lisa Su is a lot smarter than wishful thinkers of forums. She will want premium margins for her premium products.

 

[Joe Braga]

As said here, Here in Brazil the rank of best sellers CPUs is R5 2600/X, R7 2700, i5 8400 and R7 2700X and the price point here is from BRL800.00 and BRL1100.00

 

[DrMrLordX]

Vega looks a lot more like consumer cards trying to break into Datacenter than the other way around.

That is patently false. Vega20 in particular was designed for deep learning workloads. Radeon Instinct cards are PCIe4.0 CCIX-enabled implementations of Vega20, complete with IF links for IF-over-PCIe to enable an SVM compute model (as opposed to Nvidia's UVM over NVLink). Those are not consumer features, at all.

Nvidia "dominating" professional graphics sales is of no relevance. The deep learning market in particular is expanding rapidly enough to facilitate growth from multiple players. It isn't a zero-sum-gain market. AMD's Q4 2017 earnings report:

http://ir.amd.com/news-releases/new...rth-quarter-and-annual-2017-financial-results

Q3:

https://www.amd.com/en/press-releases/press-release-2017oct24

Q2:

https://www.amd.com/en/press-releases/press-release-2017jul25

Compute and Graphics drove most of the growth. Q2 saw initial sales of Ryzen. Q3 saw expanded Ryzen and initial Threadripper, plus initial sales of Vega10-based cards. Q4 saw more sales of Ryzen/TR, and more Vega10. YoY gains kept going up after Vega10 was launched.

Do you remember the RX Vega64 shortages? How do you suppose that Vega10 had such an outsized impact on YoY revenue growth when there weren't many on the market? People speculated that miners stripped the shelves clean, but I don't think sales/availability reflects that. AMD was selling most of their Vega10 dice as Pro cards.

Then AMD elected to make Vega20 a pro-only GPU. By then, it was obvious: AMD was dedicated to serving the pro market. That was where they could make the most money.

 

[PeterScott]

Nvidia "dominating" professional graphics sales is of no relevance. The deep learning market in particular is expanding rapidly enough to facilitate growth from multiple players. It isn't a zero-sum-gain market. AMD's Q4 2017 earnings report:

http://ir.amd.com/news-releases/new...rth-quarter-and-annual-2017-financial-results


Compute and Graphics drove most of the growth. Q2 saw initial sales of Ryzen. Q3 saw expanded Ryzen and initial Threadripper, plus initial sales of Vega10-based cards. Q4 saw more sales of Ryzen/TR, and more Vega10. YoY gains kept going up after Vega10 was launched.

The financials only reinforce my point, not yours. Compute and Graphics is nearly everything they do, but you will notice when they mention, the first driver is always "Desktop", not professional, not data center. AMD makes the vast majority of their money from consumer, not professional market.

Do you remember the RX Vega64 shortages? How do you suppose that Vega10 had such an outsized impact on YoY revenue growth when there weren't many on the market? People speculated that miners stripped the shelves clean, but I don't think sales/availability reflects that. AMD was selling most of their Vega10 dice as Pro cards.

Yes, I remember GPU shortages. It was mining done on commodity consumer cards. Again; Where is the evidence most Vega10 Cards were sold as Pro cards?

Then AMD elected to make Vega20 a pro-only GPU. By then, it was obvious: AMD was dedicated to serving the pro market. That was where they could make the most money.

Total Nonsense. AMD RTG makes most of it's money in consumer GPUs, and AMD is king of re-using chips. There were always going to dual purpose this chip, like they dual purposed every other "pro" GPU chip they ever sold.

 

[Kocicak]

Market share is not just about momentary sales figure, it is about how many goods from one manufacturer people really own and use. The end goal is the largest presence on the market in highest usage percentage. So even if AMD happened to outsell Intel right now, it is meaningless. They need it to last a long time and actually build costumer base to buy their future products. And for that they cannot be just a little bit better than Intel, they must CRUSH IT. They have a large gap to fill.

I will add one more thing to what I already wrote.

This is a fierce contest. If one party has an advantage, it MUST USE IT FULLY. Because if they do not, the other party, when it gets up to speed again, will profit from this unnecessary slowdown. So thinking, that somebody should WITHOUT ANY RATIONAL REASON keep prices high, even they could be lower (meaning accelerated market share increase), just because they appear low enough at this moment, would not be clever at all. BECAUSE IT IS NOT JUST ABOUT MOMENTARY SITUATION BUT ABOUT FUTURE TOO. Any unnecessary slowdown means losses now or in the future.

Mockingbird, do you understand this? Imagine a race. You just run as fast as you can, because if you slow down for no reason, the competitors will get you sooner. Is this sports analogy clear enough?

 

[OTG]

See above. it's clear that this is not AMDs strategy because we can see it in the market and because they said so themselves (not wanting to be the cheap brand).

Plus this strategy destroys their own market. Once the prices are that low, people simply won't pay more (unless it's a near monopoly, see NV pricing) and simply won't but at all until AMD/intel offer better options.

Not wanting to be the 'cheap brand' means they don't want to only be the cheap brand, not that they don't want to sell inexpensive CPUs. AMD wants people to associate them with high-performance first, so if somebody is looking for a $100 CPU, they see the AMD name and think "Hmm, AMD makes that badass $500 (or whatever) 16-core thing, I bet their 6-core is pretty good too."
Halo effect, like Intel and Nvidia have been enjoying for a long time.

If their entire lineup is priced so that Intel can't compete without losing money (or at least, hugely reduced margins), then so much the better.
That's not "being the cheap brand" it's being the brand that dominates in all categories.

I'm not sure it destroys their own market, if they're able to segment it right.
Even if the entry-level is a $99 6-core, if for only $60 more you get a 9900k-killer, then at $320 you get a 12-core threadripper-lite, etc. There's still reason to upgrade, even if the advantages are more theoretical than tangible.
It DOES kinda kill the market for dual cores outside of mobile, though.

 

[OTG]

Why aren't they already dominating the CPU commodity market with superior price/performance being recognized?

Because it's more likely that AMD has a perception/marketing/education issue, not a pricing issue.

Yes, and how do you deal with a perception issue? By offering an incrementally better product at better prices?
That's what Ryzen 1 did, and it worked pretty well, but Intel still has the high-end gaming niche giving them a halo that internet warriors won't let go of.

An 8-core that slightly beats a 9900k just means a contested top spot, a 12-core that walks away from a 9900k means there's a new top chip, and Intel can't counter it by wringing another 100mhz from 14nm and dropping the price by a hundred bucks.

If AMD had nothing but time, they could just do Intel-like incremental improvements, and perception would eventually shift.
They don't! Intel isn't a stationary target, and when they roll out their next generation products, they're almost certainly going to be *damn good*.
AMD needs products that Intel will struggle to compete with, not "the same but cheaper".
THAT makes them look like the poor man's Intel.

 

[Mockingbird]

I will add one more thing to what I already wrote.

This is a fierce contest. If one party has an advantage, it MUST USE IT FULLY. Because if they do not, the other party, when it gets up to speed again, will profit from this unnecessary slowdown. So thinking, that somebody should WITHOUT ANY RATIONAL REASON keep prices high, even they could be lower (meaning accelerated market share increase), just because they appear low enough at this moment, would not be clever at all. BECAUSE IT IS NOT JUST ABOUT MOMENTARY SITUATION BUT ABOUT FUTURE TOO. Any unnecessary slowdown means losses now or in the future. That means raising or keeping the prices as high as the market would tolerate, in order to obtain the highest profit margin and not lower the prices unless forced to do so by the other party.

I fixed it for you.




This is not OT or P&N. You cannot alter a member's quote to add your "opinion" as it was his own.


esquared
Anandtech Forum Director

 

[Kocicak]

I see. You just do not understand, that profit maximisation is not the only business objective. Business growth by sales maximisation and market share increase??? Never heard about that???