AMD Carrizo Pre-release thread

[R0H1T]

Skylake desktop wont compete with broadwell mobile.

I think it was Skylake mobile that was gonna launch first, like has been the trend recently, unless I'm missing something

 

[Fjodor2001]

I think it was Skylake mobile that was gonna launch first, like has been the trend recently, unless I'm missing something

Last I've heard it's desktop Skylake in 2015Q2, and mobile in 2015Q4 (or late Q3).

I do agree it's weird that Intel changes back and forth what is released first (mobile vs desktop). I have no clue what Intel's release strategies are these days, and they don't communicate much to the public about their intentions either anymore unfortunately.

 

[ShintaiDK]

I think it was Skylake mobile that was gonna launch first, like has been the trend recently, unless I'm missing something

Desktop first this time as shown on roadmaps.

 

[AtenRa]

Carizo will be on retail at the same time to compete against Broadwell, Carrizo-L against Cherytrail.

 

[piesquared]

Carizo will be on retail at the same time to compete against Broadwell, Carrizo-L against Cherytrail.

So Carrizo and Carrizo-L will be on the Skybridge platform, unifying the socket and the rest of the system, while OEMs will have to deal with multiple sockets and platforms for intel SKUs. intel will probably need even more contra revenue to lock AMD out of the market, but it might not matter how much money they throw at it if Skybridge drastically simplifies OEM verification and TTM.

 

[Shehriazad]

And I'm just sitting here...waiting for news about the Desktop Carrizos....*sigh*

 

[piesquared]

And I'm just sitting here...waiting for news about the Desktop Carrizos....*sigh*

Well you are in the wrong forum. This is predominantly an investor forum, and the stock market is what matters here. Mobile is as important, or more important than desktop and Carrizo/ Carrizo-L appear to have a good chance of gaining market share according to some whispers out there.

 

[Erenhardt]

If you are waiting for Carrizo news, it may not be the best place:
from AT page:

You've landed on mobile news on AnandTech. It features a collection of all of our independent mobile content and is sponsored by Intel.

I don't think intel will pay for amd coverage.

 

[frozentundra123456]

So what happened to the paid AMD section? Never heard you claiming that would lead to biased coverage.

 

[mrmt]

So Carrizo and Carrizo-L will be on the Skybridge platform, unifying the socket and the rest of the system, while OEMs will have to deal with multiple sockets and platforms for intel SKUs. intel will probably need even more contra revenue to lock AMD out of the market, but it might not matter how much money they throw at it if Skybridge drastically simplifies OEM verification and TTM.

I'm really interested in seeing what kind of compromises AMD made in order to get both Carrizo and Carrizo-L on the same socket, like, for example, the fact that we may not see anything above 35W in the Excavator line up.

I think this decision is more a defensive move, caused by the shrinking volumes of both chips which are not enough to justify the development of two new platforms, than a sound business strategy to expand market share.

Well you are in the wrong forum. This is predominantly an investor forum, and the stock market is what matters here. Mobile is as important, or more important than desktop and Carrizo/ Carrizo-L appear to have a good chance of gaining market share according to some whispers out there.

The same whispering voices that said the same about Llano, Trinity, Richland and Kaveri, or did you find a new whispering voice?

 

[piesquared]

I'm really interested in seeing what kind of compromises AMD made in order to get both Carrizo and Carrizo-L on the same socket, like, for example, the fact that we may not see anything above 35W in the Excavator line up.

I think this decision is more a defensive move, caused by the shrinking volumes of both chips which are not enough to justify the development of two new platforms, than a sound business strategy to expand market share.



The same whispering voices that said the same about Llano, Trinity, Richland and Kaveri, or did you find a new whispering voice?

I'm more interested in seeing how much contra revenue it will cost intel to keep from having to compete, speaking of defensive moves....

What about design wins? Wasn't that your main criteria previously? As mentioned, a common Skybridge platform increases design wins.

 

[mrmt]

I'm more interested in seeing how much contra revenue it will cost intel to keep from having to compete, speaking of defensive moves....

No, not a defensive move by any standards. Intel is BUYING market share, EXPANDING its footprint at the expense of its balance sheet. AMD is consolidating the two platforms because they can't stand in their own leg by their own, Carrizo and Carrizo-L will sell *less* than the sum of Kaveri and Beema, or Trinity and Jaguar. AMD will have less design wins with both Carrizo chips than it had with the combination of the previous two generations.

 

[piesquared]

No, not a defensive move by any standards. Intel is BUYING market share, EXPANDING its footprint at the expense of its balance sheet. AMD is consolidating the two platforms because they can't stand in their own leg by their own, Carrizo and Carrizo-L will sell *less* than the sum of Kaveri and Beema, or Trinity and Jaguar. AMD will have less design wins with both Carrizo chips than it had with the combination of the previous two generations.

Sounds like a lot of wishful thinking there..

It'll be interesting to see if intel's defensive contra revenue is enough to hinder competition. Likely not, as a simplified common platform with much reduced validation is an attractive proposition to OEMs.

PS, how does something stand in it's own leg?

 

[Erenhardt]

So what happened to the paid AMD section? Never heard you claiming that would lead to biased coverage.

Well guess what. After you entered amd center, there was no intel news.
Also, amd center was not the whole "Desktop" branch of the site, go figure.

No, not a defensive move by any standards. Intel is BUYING market share, EXPANDING its footprint at the expense of its balance sheet.

This is exactly what he said, but you used other words.
Paying for not having to compete on equal foot. Defend your product with $100 paycheck behind each. LOL

And you seem to defend this practice. I wonder if you would be supporting such practice if someone went to your boss and offered fat paycheck each month for firing you. Because that is basically what happens.

Hard not to notice that just a little mention of carrizo, without any actual new groundbreaking info stirred so many pots on the other side of the fence.:whiste:

 

[NTMBK]

Consolidating platforms is a great move, as it provides more flexibility to OEMs by offering a wider selection of SKUs and letting them produce a variety of models from a single laptop design.

 

[mrmt]

This is exactly what he said, but you used other words.
Paying for not having to compete on equal foot. Defend your product with $100 paycheck behind each. LOL

No, I didn't. If you can't see the difference between buying market share and narrow market focus by consolidating product lines (and leave others previously attended market brackets in the cold) I don't have anything to add. And btw, welcome to the business world. Financial muscle matters, and if you can't stomach this kind of competition, go work for an NGO.

Where's the desktop Carrizo? Where's the 45-55W SKUs? That's right, they are gone, because of the common socket they have to stay low on power.

It'll be interesting to see if intel's defensive contra revenue is enough to hinder competition. Likely not, as a simplified common platform with much reduced validation is an attractive proposition to OEMs.

A common platform will help OEM with validation and inventory management, because they will have only one AMD board to work with. But, what are the trade offs? If there was no trade off, don't you think AMD would have made that move earlier? The trade off is that both Carrizo and Carrizo-L need to share the same limitations (power consumption, connectivity for example) and the likely outcome is that it won't be as low power as a dedicated Carrizo-L platform, or it won't perform as high as a dedicated Carrizo platform could.

That will narrow AMD market scope, because of the previously mentioned trade offs, but AMD hopes it will give them a better shot in a denser part of the market spectrum. That move makes sense only if you see a cash strapped company looking to stop the market share bleeding by reducing the costs of their slightly improved market share-losing products.

Consolidating platforms is a great move, as it provides more flexibility to OEMs by offering a wider selection of SKUs and letting them produce a variety of models from a single laptop design.

Provides more flexibility for OEMs for offering of a wider selection of SKUs of the same model but narrows the TAM for the product line, because you don't have, for example, nor a barebones board for the cheapest SKUs and you don't have a high performing, neither a feature-rich board for the higher end SKUs. It's flexibility within the selected market bracket at the expense of flexibility of the entire product line.

 

[el etro]

Defending Intel supposed anti-competitive practices is a total LOL.

 

[NTMBK]

That will narrow AMD market scope, because of the previously mentioned trade offs, but AMD hopes it will give them a better shot in a denser part of the market spectrum. That move makes sense only if you see a cash strapped company looking to stop the market share bleeding by reducing the costs of their slightly improved market share-losing products.

Provides more flexibility for OEMs for offering of a wider selection of SKUs of the same model but narrows the TAM for the product line, because you don't have, for example, nor a barebones board for the cheapest SKUs and you don't have a high performing, neither a feature-rich board for the higher end SKUs. It's flexibility within the selected market bracket at the expense of flexibility of the entire product line.

Seriously? What do you expect laptop makers to differentiate with? Screen (resolution, touch or no), memory size, SSD/hard drive size, and APU performance. All of those can be altered with the same motherboard and chassis design. And if they want to make a lower cost variant motherboard, they still can't nothing stopping them. And let's face it, AMD isn't going to make it into any ultra high end laptops with Bulldozer family chips.

EDIT: Good example would be the Venue 11. Dell offers both a Broadwell and a Bay Trail version of this. If Atom and Core M shared a socket they could make a single design, but instead they had to make 2.

 

[krumme]

Consolidating platforms is a great move, as it provides more flexibility to OEMs by offering a wider selection of SKUs and letting them produce a variety of models from a single laptop design.

Absolutely. Look at the consolidation of platforms in the automobile industry even between companies like Renault and Nissan.
Platform thinking is the best for the entire industry for effectiveness and ttm ever.
And besides amd is like a 2.6b market cap company. It cant do this fast enough and need to do it even more.

 

[AtenRa]

Where's the desktop Carrizo? Where's the 45-55W SKUs? That's right, they are gone, because of the common socket they have to stay low on power.

Common socket doesn't have anything to do with TDP. Intel has SKUs spanning from 35W TDP up to 88W TDP at 22nm in a single socket 1150 for the Desktop.
AMD Richland SKUs are spanning a wide TDP range from 45W all the way up to 100W TDP with a single socket.

Also, most people have forgotten about it but there where Desktop Motherboards in the past(ASUS had a few models) that could use mobile Intel CPUs.

A common platform will help OEM with validation and inventory management, because they will have only one AMD board to work with. But, what are the trade offs? If there was no trade off, don't you think AMD would have made that move earlier? The trade off is that both Carrizo and Carrizo-L need to share the same limitations (power consumption, connectivity for example) and the likely outcome is that it won't be as low power as a dedicated Carrizo-L platform, or it won't perform as high as a dedicated Carrizo platform could.

Intel Socket 1150 has many many boards that can tolerate from 35W TDP up to more than 200W TDP(when OCed). I dont see them have any problems with that.

Also, AMDs AM3+ boards are made to sustain 95W TDPs and all the way up to 225W TDPs with a single socket. That didnt stop Motherboard makers producing different models.

That will narrow AMD market scope, because of the previously mentioned trade offs, but AMD hopes it will give them a better shot in a denser part of the market spectrum. That move makes sense only if you see a cash strapped company looking to stop the market share bleeding by reducing the costs of their slightly improved market share-losing products.

I know you are trying your best to down play the single socket configuration and make AMD look weak but you are not succeeding.

A single socket for both APUs will lower BOM because both APUs will be able to use the same platform/motherboard. It will lower validation time and money spend etc etc.
All that will make Carrizo and Carrizo-L more price competitive than any Intel APU in that category and this is what you dont like people to know.

Provides more flexibility for OEMs for offering of a wider selection of SKUs of the same model but narrows the TAM for the product line, because you don't have, for example, nor a barebones board for the cheapest SKUs and you don't have a high performing, neither a feature-rich board for the higher end SKUs. It's flexibility within the selected market bracket at the expense of flexibility of the entire product line.

Having a single socket doesnt prohibits you from having multiple boards, AMD AM3+ and Intel Socket 1150 is a nice paradigm you specifically trying to avoid mentioning here, i wonder why

 

[mrmt]

Seriously? What do you expect laptop makers to differentiate with? Screen (resolution, touch or no), memory size, SSD/hard drive size, and APU performance. All of those can be altered with the same motherboard and chassis design.

Battery life and raw performance are both impacted by the MB choice, because it does impact power consumption and offers a ceiling on how much power you can provide to the CPU. Go too high on power consumption ceiling and you impact low power performance, the opposite is true.

EDIT: Good example would be the Venue 11. Dell offers both a Broadwell and a Bay Trail version of this. If Atom and Core M shared a socket they could make a single design, but instead they had to make 2.

Isn't that what I was saying all the time? For a given market bracket, the one that is the focus of AMD strategy, it will do ok, but go to the high end or at least high performance notebooks and Carrizo becomes a no-go, and I think we can't expect a Carrizo tablet, Carrizo barebones or a Carrizo Surface for the same reasons. AMD got more focused in a given market bracket at the expense of other niches that could be served by its processors.

 

[NTMBK]

Battery life and raw performance are both impacted by the MB choice, because it does impact power consumption and offers a ceiling on how much power you can provide to the CPU. Go too high on power consumption ceiling and you impact low power performance, the opposite is true.

BS. These are both full SoCs with no external southbridge; only noticeable MB power draw will be stuff like wifi modem which are common across all models.

 

[AtenRa]

Battery life and raw performance are both impacted by the MB choice, because it does impact power consumption and offers a ceiling on how much power you can provide to the CPU. Go too high on power consumption ceiling and you impact low power performance, the opposite is true.

Isn't that what I was saying all the time? For a given market bracket, the one that is the focus of AMD strategy, it will do ok, but go to the high end or at least high performance notebooks and Carrizo becomes a no-go, and I think we can't expect a Carrizo tablet, Carrizo barebones or a Carrizo Surface for the same reasons. AMD got more focused in a given market bracket at the expense of other niches that could be served by its processors.

You can have the same board with fewer VRMs, Caps etc etc for the lower TDP SKUs and increase the VRMs, caps etc for the higher TDP SKUs. Much like AMD and NVIDIA is doing with Discrete GPUs. Same board, fewer parts, less validation etc etc = lower price.

 

[mrmt]

BS. These are both full SoCs with no external southbridge; only noticeable MB power draw will be stuff like wifi modem which are common across all models.

Be my guest. Let's see around Q414 what's left of AMD mobile market share. Not much I would say.

 

[NTMBK]

Isn't that what I was saying all the time? For a given market bracket, the one that is the focus of AMD strategy, it will do ok, but go to the high end or at least high performance notebooks and Carrizo becomes a no-go, and I think we can't expect a Carrizo tablet, Carrizo barebones or a Carrizo Surface for the same reasons. AMD got more focused in a given market bracket at the expense of other niches that could be served by its processors.

Carrizo would never hit high end laptops even without this, and we both know it, and AMD has Mullins and Nolan aimed at the tablet market. Kaveri tried to hit premium laptops and didn't exactly do too hot.