AMD presents future plans at investor meetings

[csbin]

http://www.planet3dnow.de/cgi-bin/newspub/viewnews.cgi?id=1364293981

Now one might think that AMD will liquidate the old PC market segment accordingly, rumors about the end of the Bulldozer architecture and the FX series did have the round, but at least partially can all clear given, because AMD shows in the presentation the familiar bulldozers architecture planning with Steamroller and Excavator

 

[inf64]

Maybe g-translate?
Anyway, it's interesting they still list SR core as a 2013 product on the slide that clearly talks about Opteron line-up. Strange stuff. Maybe they are referring to the core itself and that it will launch as an "APU" first or maybe they are planning to push 3M/6C APUs as Opterons even this year, which is not impossible but it's less likely to happen. Infrastructure is there, new SR APUs *should* work on FM2 socket even though they(Kaveri) do support various memory standards (akin to first Denebs that were AM2+ compatible but memory controller had already built-in support for DDR3).

 

[NTMBK]

And they still have Steamroller labelled for 2013... interesting.

 

[csbin]

Maybe g-translate?
Anyway, it's interesting they still list SR core as a 2013 product on the slide that clearly talks about Opteron line-up. Strange stuff. Maybe they are referring to the core itself and that it will launch as an "APU" first or maybe they are planning to push 3M/6C APUs as Opterons even this year, which is not impossible but it's less likely to happen. Infrastructure is there, new SR APUs *should* work on FM2 socket even though they(Kaveri) do support various memory standards (akin to first Denebs that were AM2+ compatible but memory controller had already built-in support for DDR3).

right!​

 

[csbin]

And they still have Steamroller labelled for 2013... interesting.

Server^_^

 

[Dribble]

tbh slide reads as:
-we get 5% revenue from consoles today.
-we expect to get 20% of revenue from new consoles by end of the year.
-we will then magically be able to get half our revenue from embedded devices in the future.

Being as the only thing that has really ever given embedded revenue so far is consoles and that certainly won't bring in 50% revenues (unless AMD shrink a lot) do they really believe that slide or is it just marketing bull?

 

[SlowSpyder]

And they still have Steamroller labelled for 2013... interesting.

That's a pretty old slide, though. I've seen that one for a while now.

Here's an atrcle that is 10 months old that has it: http://www.anandtech.com/show/5057/the-bulldozer-aftermath-delving-even-deeper

 

[inf64]

That's a pretty old slide, though. I've seen that one for a while now.

Here's an atrcle that is 10 months old that has it: http://www.anandtech.com/show/5057/the-bulldozer-aftermath-delving-even-deeper

Not the same slide. New slide has 2013 listed ABOVE SR core in the graph,old one doesn't.

 

[strata8]

tbh slide reads as:
-we get 5% revenue from consoles today.
-we expect to get 20% of revenue from new consoles by end of the year.
-we will then magically be able to get half our revenue from embedded devices in the future.

Being as the only thing that has really ever given embedded revenue so far is consoles and that certainly won't bring in 50% revenues (unless AMD shrink a lot) do they really believe that slide or is it just marketing bull?

You did a nice job ignoring the 'dense server' and 'ultraportable/low-power' dot points within that 50% figure.

Here's the full PDF of the presentation if anyone wants to take a look:
http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MTc1OTU5fENoaWxkSUQ9LTF8VHlwZT0z&t=1

 

[Idontcare]

http://translate.googleusercontent....37&w=o&usg=ALkJrhiaRl0_Qc8XaAZ0vsmSxraq5pPX8Q

"Embedded/Semi-Custom on-track to be 20% of revenue by Q4-13"

I understand we are supposed to read that and just assume that is a good thing, but is it really?

It is easy to get embedded revenue to be 20% of overall revenue, just keep shrinking the revenue you get from "traditional computing" and voila, your embedded revenue as a percentage of overall revenue is growing!

^ I am not surprised they continue to show this, the quickest way to kill your piledriver-based Opteron server sales today is to tell the customer their upgrade path is already dead because the product line has been EOL'ed internally.

It is the same reason why HP took Oracle to court over Oracle's outing of the fact that HP knew Itanium's roadmap was limited and counting down but that HP was intentionally withholding that info from its customers.

Considering how much AMD was willing to continue the marketing game with bulldozer right up to the time of its final release - including their server rep JFAMD posting blogs and video about its performance going to 11 - I am not at all surprised to see AMD continuing on in the tradition of treating their customers like idiots and fools.

 

[inf64]

Upgrade path? Really IDC? AMD has been more than generous towards their clients who bought 6100 MC systems. They got 2 generations worth of upgrades on the same platform. You are asking for 3rd generation on a platform that has no room left in it to grow. If you look at intel, they have no problems upgrading their sockets whenever they feel like they have to.
If new SR core requires new platform in order to perform better then it's the best way to go even for consumers. They would get a platform that will, similarly to initial MC one, be ready for at least one if not 2 next gen. product upgrades .

BTW PD based opterons(6300) are not doing bad at all. AT had a good article that exposes their strengths and weaknesses. They are still not the "ultimate alternative" to Xeon but they do present a viable alternative. You can also check spec.org for opteron 6300 and xeon SB-E(16T) results and see what results say.

 

[bgt]

And they still have Steamroller labelled for 2013... interesting.

Great, my next 8 core CPU:biggrin:

 

[Idontcare]

Upgrade path? Really IDC? AMD has been more than generous towards their clients who bought 6100 MC systems. They got 2 generations worth of upgrades on the same platform. You are asking for 3rd generation on a platform that has no room left in it to grow. If you look at intel, they have no problems upgrading their sockets whenever they feel like they have to.
If new SR core requires new platform in order to perform better then it's the best way to go even for consumers. They would get a platform that will, similarly to initial MC one, be ready for at least one if not 2 next gen. product upgrades .

BTW PD based opterons(6300) are not doing bad at all. AT had a good article that exposes their strengths and weaknesses. They are still not the "ultimate alternative" to Xeon but they do present a viable alternative. You can also check spec.org for opteron 6300 and xeon SB-E(16T) results and see what results say.

That's not the kind of upgrade I am talking about. The server guys in the audience know well what I am referring to and how it impacts their purchasing decisions when it comes to the confidence they have in their supplier's maintaining a continuing presence in same marketspace for after-sales support and so forth.

There is a reason AMD's server market share is as low as it is, don't fault me for recognizing that reality nor the reality of why companies have a vested interest in clouding the future of their product lineups.

If mindshare and market momentum were not a factor in motivating sales then marketing would not have invested as much as it did into JFAMD's compensation package to have him busy himself making blogs and PR videos for a product that wasn't purchasable at the time.

You act like what I am touching on is impossible or irrelevant, AMD's own spending says otherwise. If there wasn't a psychological need for customers to think that AMD will be in the big-core server market indefinitely then there would not be a need to show such roadmaps to analysts and customers in the first place.

AMD is trying to avoid Nokia's fate, Nokia announced Symbian support was going away in the future only to watch market demand for their still supported Symbian phones completely dry up in the channel. Nokia did right by the customer, letting them know the lay of the land, but did not do right by their shareholders who got socked in the gut. AMD is doing the opposite, and good for them.

 

[frozentundra123456]

Hopefully for AMD a lot of that embedded revenue is going to come from tablets as well as consoles. However, I have been questioning a lot in my mind how the consoles will affect APU sales. We dont know about xbox yet but for sure the PS4 will be much, much superior to Trinity/Richland for gaming. Unless Kaveri ups the bar a lot, it will be trailing the consoles badly, and I would not see much place for it as a gaming chip except in a mobile platform. And before anyone gets too upset, obviously the same would apply to any intel igp, even Haswell, except again, in a mobile platform.

 

[NTMBK]

Hopefully for AMD a lot of that embedded revenue is going to come from tablets as well as consoles. However, I have been questioning a lot in my mind how the consoles will affect APU sales. We dont know about xbox yet but for sure the PS4 will be much, much superior to Trinity/Richland for gaming. Unless Kaveri ups the bar a lot, it will be trailing the consoles badly, and I would not see much place for it as a gaming chip except in a mobile platform. And before anyone gets too upset, obviously the same would apply to any intel igp, even Haswell, except again, in a mobile platform.

The main issue is the memory bandwidth. They can't up the graphics that much more without some big push on memory. They need GDDR5, DDR4, EDRAM, or triple/quad channel DDR3, or any more powerful graphics cores are going to be crippled. Intel have beaten them to the punch with Haswell GT3e, and hopefully AMD have something up their sleeves (there have been GDDR5 rumours for Kaveri).

 

[Phynaz]

Here's the full PDF of the presentation if anyone wants to take a look:
http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MTc1OTU5fENoaWxkSUQ9LTF8VHlwZT0z&t=1

Interesting...In another thread there are statements that AMD doesn't incur the costs of node development anymore. But look at at slide 19 and you see that AMD is reimbursing GloFo for R&D.

 

[Insert_Nickname]

Intel have beaten them to the punch with Haswell GT3e, and hopefully AMD have something up their sleeves (there have been GDDR5 rumours for Kaveri).

Found this. Looks like GT3e will only be available in core i7 form...

http://www.xbitlabs.com/news/graphi...l_Haswell_Graphics_Options_for_Notebooks.html

 

[Tsavo]

Interesting...In another thread there are statements that AMD doesn't incur the costs of node development anymore. But look at at slide 19 and you see that AMD is reimbursing GloFo for R&D.

And paying for unmade wafers. Probably also paying for sack lunches and overdue parking tickets. And Friday pizza parties. And executive orgies. Probably a few 64+ sail yachts, too.

GF showed me a vending machine the executives put in all the women's restrooms. Filled with cheap pantyhose and bad tampons. I'd swear on the Northern Star that they had an AMD logo on them...I should have taken a picture.

This was in a hospital. :hmm: Pantyhose as common garb went out in the 70's....as for tampons....I don't know what a bad tampon is, so I rely on my GF's judgement on this.

She quit the hospital job and moved to a small clinic. That was a year ago. More money. I guess, and less hours.

 

[NTMBK]

Found this. Looks like GT3e will only be available in core i7 form...

http://www.xbitlabs.com/news/graphi...l_Haswell_Graphics_Options_for_Notebooks.html

That's a silly move in my eyes. Valve are about to kick off Steambox, and if that takes off then world + dog will want powerful gaming APUs. If Intel made desktop socketable GT3e chips, they could make for really powerful Mini-ITX machines.

 

[blckgrffn]

T
AMD is trying to avoid Nokia's fate, Nokia announced Symbian support was going away in the future only to watch market demand for their still supported Symbian phones completely dry up in the channel. Nokia did right by the customer, letting them know the lay of the land, but did not do right by their shareholders who got socked in the gut. AMD is doing the opposite, and good for them.

When I read your first post I got something totally different from what you wrote. This paragraph of clarification really helps me figure out what your were pointing out there.

It's still difficult to grasp how AMD has put themselves into this position.

 

[AtenRa]

Interesting...In another thread there are statements that AMD doesn't incur the costs of node development anymore. But look at at slide 19 and you see that AMD is reimbursing GloFo for R&D.

Moving to standard 28nm process technology and significantly reducing reimbursements to GF for future research and development costs, a savings of ~$20M per quarter during the next several years

If im not mistaken, it says they saving 20M per quarter(80M per year) by cutting off spending's for future R&D (SOI etc). It is exactly what we were talking in the other thread.

 

[AtenRa]

It's still difficult to grasp how AMD has put themselves into this position.

What position ??? As IDC have already said and what Intel and others do, AMD clearly shows that SR and Excavator are on the roadmap and Customers can still plan accordingly.

 

[blckgrffn]

What position ??? As IDC have already said and what Intel and others do, AMD clearly shows that SR and Excavator are on the roadmap and Customers can still plan accordingly.

The (extremely) disadvantaged foundry position they are in? The competitive desktop chip business where they founder about with the whole FX line? The server business where their market share and mind share - once seeming so promising with density and virtualization - diminished and bordering on irrelevancy?

From 5-6 years ago to now, is this where we would have anticipated AMD to be? Shrinking? Specifically I am talking about their Desktop/Workstation/Server CPU - which is all they really had at that point.

It was likely unrealistic that they go toe to toe with Intel and that they needed to carve out some niches that Trinity, Bobcat and upcoming smaller chips hold. These niches are going to be less lucrative than what they were targeting before by their very existence, if they were big enough and profitable enough, Intel would fill them. If Intel has unbridled dominance in the more lucrative areas for long enough, how long before they come for those niches to keep up volume?

Trust me, I like AMD and their "underdog" but somehow competitive position. I helped Microcenter sell three more FX bundles over the weekend through recommendations. But I can tell they have fallen. In another five years, will they be a one trick (custom embedded x86) pony? Maybe.

If you disagree, that's fine. It's the Internet.

 

[AtenRa]

@ blckgrffn

Well we were talking about the SR and Excavator on the roadmap slide, i thought you were talking about that not in general.

But i will like to point out that Intel is investing more and more die space for the iGPU than the CPU. Things in the x86 market will not be the same coming the near future. AMD still has the edge and the lead in that field, they only need a little help from GloFo(coming as better and on time node shrinks) to be able to sustain it.

 

[lagokc]

The main issue is the memory bandwidth. They can't up the graphics that much more without some big push on memory. They need GDDR5, DDR4, EDRAM, or triple/quad channel DDR3, or any more powerful graphics cores are going to be crippled. Intel have beaten them to the punch with Haswell GT3e, and hopefully AMD have something up their sleeves (there have been GDDR5 rumours for Kaveri).

If nVidia's Volta will have stacked on chip in 2016 then it's likely AMD will be pushing for the same thing. Imagine an APU with several GB of 1TB/s memory shared between the CPU/GPU. That would directly solve any bandwidth problems and allow AMD to put absurd amounts of GPU performance on chip.

Assuming there isn't some massive overheating issue and assuming AMD can survive that long...