AMD Zen - Key Dates and Information

[Atari2600]

Since its never been a product or SKU from AMD you expect what again?

It doesn't exactly take much brainpower to work out that evidence saying Zen+Vega APU has a 512 or 1024 bit memory controller would be a good start.

C'mon you can do better than this, your badly letting down trolls everywhere with your dismal performance so far.

 

[ShintaiDK]

It doesn't exactly take much brainpower to work out that evidence saying Zen+Vega APU has a 512 or 1024 bit memory controller would be a good start.

Where is this evidence? Its not your vague 2-3 year old slides where one may even be fake.

I know the dream of HBM on APUs have been a much desired thing for a group of people for a real long time. But again, reality and fantasy usually never mix.

 

[Atari2600]

Where is this evidence?

I am expecting YOU to provide the evidence.

YOU are saying the APU won't have a 2048 bit mem controller.
I have provided evidence it will.
YOU have yet to provide evidence to the contrary.


Very simple to understand. Even for simpleton trolls.

Insulting other members is not allowed
Markfw
Anandtech Moderator

 

[Glo.]

Where is the evidence that is otherwise, Shintai? Just because you claim it to be that way?

http://videocardz.com/62250/amd-vega10-and-vega11-gpus-spotted-in-opencl-driver

Vega is the same family as Greenland, and Raven Ridge. Therefore Raven Ridge APUs have HBM2 support. If they have HBM2 support, the CPU also needs to have HBM2 support.

Then we have slides showing that Greenland is an APU with 2048 bit controller. Vega has only 2048 bit memory controller for HBM2.

Because HBM2 is inherent part of the architecture. I'm guessing you will only believe after you will see the HBM2 APUs. But first, try to prove that your point is more valid than others.

 

[zinfamous]

Guys, these threads tend to be more productive if you just ignore Shintai.

 

[positivedoppler]

You can try and change the goalpost as much as you want, there still isn't any KNL products without the 6 channel DRAM and nobody else is even close to have similar capabilities.

Vega 20 will be 2 years after Nvidia just for GPU peer to peer communications. And around 2 years behind Intel as well with KNL on Omnipath.

Why are you trying to, twist this thread into an AMD vs Nvidia?

 

[rainy]

Why are you trying to, twist this thread into an AMD vs Nvidia?

Because effectively it's what he doing on AT forum for the last a couple years - trolling/bashing and looking for flame war.

Btw, he must be very busy lately, over 2200 posts in just a bit over 6 months:
https://hardforum.com/members/shintai.293808/

 

[DrMrLordX]

lets say they ship 20million units this year(could be twice that really)

Let's not:

http://www.xbitlabs.com/news/cpu/display/20060807143609.html

Bear in mind that the PC market has shrunk considerably since '06, and that '06 was the last time that AMD had a competitive product (granted that ended in May/June '06).

The expected 4c, 8c, and possible 6c Zen products coming from AMD in . . . February? are roughly analogous to the x2 and FX chips aimed at the socket AM2 platform in '06.

I would expect AMD to sell maybe 10 million Zen units, tops. Probably less. And that includes China. AMD will not launch any "budget" options based on Zen until Q3 2017 at the earliest when they start selling Raven Ridge. You will absolutely not see $99 Summit Ridge among the initial product offering, period. There is not enough demand for that to make sense, and AMD doesn't have the marketing know-how to throw off their reputation as an also-ran anyway.

I think both its very difficult to make people upgrade and take new sales in a mature if not declining desktop market

Bing bing bing! You win.

Stating that people don't need an octa is opinion and what matters more is if they WANT one. We are not talking about the average consumer but folks that care a bit more.

So are we looking at less than one million Zen units shipped this year? Sounds like you're saying just that . . .

AMD really can't afford to price it out of range or give Intel a break.

All they have to do is charge less than Intel for +/-10% performance. Thanks to the HEDT line, that'll be easy for their 8c/16t offerings. And it's not like Intel's 4c/8t chips are cheap either.

 

[dahorns]

It is really easy to lower prices to drive volume. It is really hard to raise them and get people to pay. AMD has already done its time as the budget brand. I'm betting it prefers not to go back to it. For this reason initial prices will be higher than expectations in this thread, i.e. $500+ for the 8c parts.

 

[rainy]

Let's not:
Bear in mind that the PC market has shrunk considerably since '06, and that '06 was the last time that AMD had a competitive product (granted that ended in May/June '06).

It's really surprising for me that so many people are saying that AMD hadn't competitive CPU in the last decade.
What was wrong with Phenom II X4/X6 (2009/2010) - in fact, it was a very decent processor.

 

[swilli89]

Guys, these threads tend to be more productive if you just ignore Shintai.

Seriously, just ignore/don't reply to his posts and he will go away.

 

[krumme]

Let's not:

http://www.xbitlabs.com/news/cpu/display/20060807143609.html

Bear in mind that the PC market has shrunk considerably since '06, and that '06 was the last time that AMD had a competitive product (granted that ended in May/June '06).

The expected 4c, 8c, and possible 6c Zen products coming from AMD in . . . February? are roughly analogous to the x2 and FX chips aimed at the socket AM2 platform in '06.

I would expect AMD to sell maybe 10 million Zen units, tops. Probably less. And that includes China. AMD will not launch any "budget" options based on Zen until Q3 2017 at the earliest when they start selling Raven Ridge. You will absolutely not see $99 Summit Ridge among the initial product offering, period. There is not enough demand for that to make sense, and AMD doesn't have the marketing know-how to throw off their reputation as an also-ran anyway.



Bing bing bing! You win.



So are we looking at less than one million Zen units shipped this year? Sounds like you're saying just that . . .




All they have to do is charge less than Intel for +/-10% performance. Thanks to the HEDT line, that'll be easy for their 8c/16t offerings. And it's not like Intel's 4c/8t chips are cheap either.

Its not about performance but segments.
There is no consumer market for cpu above 500 usd worth talking about. Its certainly less than 0.5M this year. And certainly not for amd. You mention the problem yourself.

You can say amd can just use the 8c for servers and then use 4 and 6 c for desktop. Thats a viable strategy if servers is expected to rock 10% plus this year. Then it just looks weird to me to name it like the core line with sr6 vs a i7. Mehhh. It simply doesnt add up.

 

[f2bnp]

It's really surprising for me that so many people are saying that AMD hadn't competitive CPU in the last decade.
What was wrong with Phenom II X4/X6 (2009/2010) - in fact, it was a very decent processor.

Or how about Phenom II x3 720 unlocking to x4 and slight overclocking to 3-3.2GHz ?

 

[inf64]

I doubt top 8C/16T Ryzen bundled with good AIO WC will cost more than 500$. Next down the line 8C SKU then will be around 80-85% of that price, so 400-425$. It follows then that top 6C/12T part will be around ~340-360$ and top 4C/8T part around 230$.

So AMD will basically:
-"kill" higher end 8C/6C BDW-E line with its 16T SKUs (~equal performance, a lot cheaper, good OCing margin )
-"kill" lower end 6C BDW-E and top mainstream i7s with 6C/12T Ryzen SKUs( ~better or equal performance, cheaper,good OCing margin )
-undercut KL i5 range with 4C/8T(~better or equal performance, cheaper, unlocked) and 4C/4T(~somewhat slower but also unlocked and way cheaper).

 

[Atari2600]

Its not about performance but segments.
There is no consumer market for cpu above 500 usd worth talking about.

The top line Zen will be in the >500 bracket. Possibly even the top two in the Zen line > 500 USD.

On newegg - 2nd in the 2015 revenue was a CPU at >1000 USD. Which would make it easily the most profitable CPU sold in 2015 (on newegg).

Now obviously OEMs etc will average much further down the performance & cost chart meaning the 2nd highest revenue CPU sold by Intel for consumers in 2015 was not the 5960X, but newegg are not insubstantial, and AMD are primarily launching Zen at folks who would peruse such sites.

https://blog.neweggbusiness.com/components/best-selling-cpus-of-2015/

 

[SpaceBeer]

Small thoughts and some calculations

Let's say AMD is going to release Ryzen CPUs in 2nd half of February. In case they want to ship them from East Asia, they should already be ready in order to have them on shelves in USA and EU (it takes 3-6 weeks by ship). We don't know size of the CPU box, which depends on cooler (Wraith or Near-Silent 95W). I think you could fit 700-800 pcs on pallet, which would be ~20.000 in shipping container. If I followed statistics correctly, around 300.000 PCs is sold per day in these markets. Ryzen might be interesting for let's say ~5-10% of them, which is ~20.000 per day. And no one is sending one container every day. So they already have them ready, or could use air freight which increases price a lot. I would say $20-50 per CPU, depending on CPU price and quantity transported.

 

[jackstar7]

How about you do the math and show us how their costs really are and why very high margins would not be sufficient? Or how pricing it high would generate the most profits.
It's the same cost as a 4 cores APU, lower if you consider the GPU IP.
Octa cores at 249-349$ or w/e is giving it away? They could sell it at 99$ and still make a billion in gross profits this year from this platform.
You make empty claims based on nothing, you haven't provided any relevant argument about pricing,margins, volumes or any other additional relevant factors. Bring something of value to the conversation.

Your lack of self-awareness makes this conversation moot.

Talk about empty claims. Yeesh.

 

[krumme]

The top line Zen will be in the >500 bracket. Possibly even the top two in the Zen line > 500 USD.

On newegg - 2nd in the 2015 revenue was a CPU at >1000 USD. Which would make it easily the most profitable CPU sold in 2015 (on newegg).

Now obviously OEMs etc will average much further down the performance & cost chart meaning the 2nd highest revenue CPU sold by Intel for consumers in 2015 was not the 5960X, but newegg are not insubstantial, and AMD are primarily launching Zen at folks who would peruse such sites.

https://blog.neweggbusiness.com/components/best-selling-cpus-of-2015/

There sure is a huge marketing effect if 8c zen beats a bwe 8c over most of the loads and thereby is declared a champion and the fastest cpu in reviews. Then highest bin will go for sky high shiny prices and the 2 and perhaps third in line will tend to go over the 500 usd line also.
That kind of performance victory remains to be seen. I doubt it.
You guys also have to adress the elephant in the room. The naming. Give me an explanation for that?

 

[inf64]

You guys also have to adress the elephant in the room. The naming. Give me an explanation for that?

Can you clarify what do you mean by naming?

 

[krumme]

Can you clarify what do you mean by naming?

That zen should be named sr3 sr5 sr7 like the core i3 i5 i7 line.
https://community.amd.com/thread/207944
Dont know if there is a confirmation of this?

 

[lopri]

It's not going to be Athlon, Sempron, FX. Maybe giving the core count emphasis is in order?

X8-xx00 (8C/16T)
X6-xx00 (6C/12T)
X4-xx00 (4C/84T)

It also has a virtue of having a higher number than Intel CPUs. (i3i5/i7)

 

[Doom2pro]

Somebody PLEASE create a Ryzen Price Speculation thread.... *sigh*

 

[inf64]

That zen should be named sr3 sr5 sr7 like the core i3 i5 i7 line.
https://community.amd.com/thread/207944
Dont know if there is a confirmation of this?

I don't see why that is strange? Intel has i7 SKU named 6950 that is 10C/20T. Why can't AMD have SR7 SKU that is 8C/16T?

 

[Atari2600]

That zen should be named sr3 sr5 sr7 like the core i3 i5 i7 line.
https://community.amd.com/thread/207944
Dont know if there is a confirmation of this?

**If** the SR3 starts at 220 USD and is a low-binned 4C part, surely that in itself is enough to suggest the highest binned 8C16T part is >500 USD.

Otherwise its gonna be very messy with overlaps surely?

Could they really go something like the below?

4C8T -- low bin: 220
4C8T -- med bin: 250
4C8T -- high bin: 280

6C12T -- low bin: 300
6C12T -- med bin: 340
6C12T -- high bin: 380

8C16T -- low bin: 400
8C16T -- med bin: 450
8C16T -- high bin: 500 (ignoring that the halo part is usually a ludicrous step above next cpu)

 

[krumme]

I don't see why that is strange? Intel has i7 SKU named 6950 that is 10C/20T. Why can't AMD have SR7 SKU that is 8C/16T?

You are right in that sense. I just interprete it against the non hedt line. I might be wrong. I just dont see it as a interesting segment selling 600usd cpu so therefore interpreted the naming as they were gunning for up till i7 4c desktop.