AMD Reportedly Scraps 28 nm APUs at GlobalFoundries

[Guest]

The announcements made at the recent "worldcast" held by AMD's new CEO Rory Read have been kept a good secret with very little information trickling out to the public.

AMD Reportedly Scraps 28 nm APUs at GlobalFoundries : Read more

 

[Guest]

He needs to bring the CPU business back on track by throwing ending Bulldozer quickly and developing a new CPU architecture. Steal from Intel!

 

[fazers_on_stun]

Well I guess this puts the AMD community reporter's statement that everything between AMD and GF is hunkey-dorey to rest..

 

[freggo]

AMD will ALWAYS be behind; as Intel effectively controls the technology and the market. Naturally, AMD can just play 'follow the leader'. So if Intel comes out with something new, surprise... AMD will be behind and catching up; again. Frustrating position to be in.

 

[joytech22]

Pity..

AMD is in some strife, I really REALLY hope they make a comeback. There needs to be competition in the market in the high-end segment.

I'm planning on getting a FX8120/FX8150 for a small home VM server and just to generally play around with to find it's strong and weak spots. Oh and to help out team Green.

 

[jdamon113]

amd should do the maunfacturing itself. if it goes to tsmc, I will never suggest one and in my network we will never attempt a amd cpu.
This dushbag needs to get a clue. Do the work youself. stop relying on others to do it for you.

 

[Guest]

AMD should be thinking more along the line of 22nm if they are thinking starting from scratch, though i do not know if TMSC could provide them with a solution.

 

[kyuuketsuki]

[citation][nom]freggo[/nom]AMD will ALWAYS be behind; as Intel effectively controls the technology and the market. Naturally, AMD can just play 'follow the leader'. So if Intel comes out with something new, surprise... AMD will be behind and catching up; again. Frustrating position to be in.[/citation]
AMD may be stuck playing catch-up as far as CPU performance is concerned, but Intel will likely be stuck playing catch-up to AMD's APUs. I'd still take a Llano laptop/ultrabook over any Intel solution any day. People don't seem to get that CPU performance is at a "good enough" level nowadays and most consumers won't notice much, if any, difference between going with an expensive Intel Core-whatever or a cheaper AMD solution unless they are gamers that must play on the highest quality settings or actually use their computers for computationally intensive tasks fairly often. The vast majority of the market does not fit into those categories.

 

[supall]

I'm sorry, but I don't just hop on the "AMD Hate" bandwagon that seems to have swept everyone else's feet. Its great that AMD is buckling down and is trying to make changes to regain its composure. I don't understand why you have people wanting to see the downfall of a company that gives us something other than "Intel". If AMD focuses more on its APUs we can see them overtaking the low-end/mid-range market, which has the larger and more profitable base. Yeah, their Bulldozer didn't sweep Intel, but it doesn't mean they are failing. If I recall correctly, their APUs are selling like hotcakes and Bulldozer is decent enough to be considered in a build.

Just like I enjoy Android but don't want to see Apple fail, I don't want to see AMD fail. I want them to succeed and force innovation. AMD brings something different and that's what we need in today's world.

 

[TeraMedia]

IIRC, TSMC had something like a 4% success rate with early high-end GPUs from AMD/ATi and NVidia. 96% went to the bin.

If AMD thinks that this is better than what GF can do, then GF is in a very scary place right now.

FWIW, I honestly believe that chip designers and manufacturers should be different. While there may not be a higher-margin device than an Intel chip right now, what if there were? And what if a third party wanted to use Intel's fab to make it, because that fab was the only one capable of doing so? If the price is right, and the capacity is there, why should Intel be the only one allowed to use it?

 

[dogman_1234]

I hope Roy takes AMD of GloFo's list and shows GF what their IQ is.

 

[Parsian]

That is good. They just need to focus on high yield for APUs specially on mobile market, currently that is their strongest point on consumer market. Coupling that with OpenCL to strengthen their grip not only in media acceleration but also in HPC market.

(I wonder if it is feasible/possible for AMD to get into ARM as well?)

AMD needs to compete with nVidia in HPC. They have the capacity, they need to raise their support and development bar. Why they are not competing in Server GPUs as effective as nVidia??? Their Server CPU and GPU solution -> cheaper cost for HPC from on vendor??

I personally think, the new guy is very competent. AMD is behind and that is a fact, they need to focus on opportunities that are opening/closing before it becomes too late.

And as for desktop, I think when it comes to consumer media needs (gaming including) GPU is playing much effective role than CPU. GPGPU is taking over the most CPU intensive tasks. I am happy with X6 Phenom and I dont see myself upgrading soon.

lastly, for AMD to stay competitive in X86, they need to throw money in R&D and prioritize server competition and MARKETING!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

 

[wiyosaya]

In my opinion, this is a good move for AMD. A better move might be to make the chips themselves, however, that is an expensive proposition if they do not already own the equipment. If AMD can bring it in house, I think they should.

I worked for a global photographic leader, and, IMHO, their main failing was that they wanted to buy everything off the shelf, and if they could not, they expected that exactly what they needed would magically appear on the market when they needed it so that they could buy it. In my opinion, that was a big failing of theirs since they had the intellectual know-how to make everything they needed themselves. It is not always cheaper to outsource, and it seems that in many cases it is cheaper for a company to develop their needed capabilities themselves rather than outsource; start-up costs are high, however, the payoff will come if the venture is successful - and - perhaps most importantly - a company that positions themselves as such will be at no one's mercy.

 

[Cazalan]

That's a fairly big set back. AMD will have to re-tapeout their APU for TSMC, which could take 6 months to a year or more.

 

[Onus]

While this may look like an amputation, sometimes that's what the patient needs. This CEO seems to have a clue. I hope for the sake of competition and lower prices that he is successful.

 

[fulle]

*sigh*
@supall
Nobody hates AMD, they're just worried about all the bad decision making. You say that Llano is selling like hotcakes, but they had a lot of fab issues, and didn't pump out a whole lot of supply on those, leading to some frustrated partners, and a lot of lost opportunity. You say Bulldozer is "decent enough to be considered in a build", but I disagree. Looking at Bulldozer's performance, efficiency, and price, NO, it's NOT, worth considering in almost ANY build.

Which has me really concerned about Trinity. AMD is reporting 20-30% gains vs Llano, but Ivy Brige has made up a LOT of ground on the GPU. I think that AMD will still have the GPU performance advantage, but Intel's integrated graphics will be "good enough" for a lot of users, and the CPU performance and power consumption numbers should put Trinity to shame badly. So badly, in fact, that I worry that Trinity is going to lose the market for things like HTPCs to Ivy Bridge stuff, as well as get it's ass kicked in the mobile sector.

As far as Apple goes... No, actually, I think it's legitimate to hope Apple fails. They're a negative company when it comes to innovation, basically patent blocking eveyone, suing business partners, and doing their best to make the most closed environment possible for their users, so that they can have as much control over the content possible, and leverage unfair prices. They're one of the most evil tech companies around right now, to be perfectly honest.

So, yeah. Pretty much disagree with you on all points, except for the general "I wish AMD would do better, since we need them to compete with Intel." To which I totally agree. But, I'm not going to go as far as buy a POS FX8120 or something. If I'm going to go that far, I'll buy an Ivy Bridge, and then mail AMD 20 bucks in a envelope with a note that says "PLZ stop sucking, you just made me buy Intel again".

 

[robwright]

I'm confused by the ExtremeTech story. The author cites "independent sources" that claim the 28-nm APUs are in serious trouble and "likely" would not see the light of day. I read that as "Outside sources say these chips are in serious trouble and will probably be cancelled" rather than "AMD has cancelled them, and they are as dead as fried chicken."

But the next sentence is also confusing: "AMD reportedly finalized the decision to cancel..."

Uh, reported by who? The independent sources? Or AMD? Or a third-party news outlet not mentioned in the story? If that information was received by the independent sources....then why are they saying the products "LIKELY won't see the light of day"? If these sources knew that AMD had finalized the decision to cancel the products, well, then there's no "likely" about it. Story just seems a little....unclear to me.

 

[_SirO_]

I call it a bluff.
While he might be frustrated and tired of working with GlobalFoundries, the impact of such a decision is just too big. He is probably going after some kind of "discount" or compensation from GlobalFoundries, while also indirectly telling them to get their act together and deliver.

 

[ta152h]

Real men have fabs.

AMD is a bitch now.

 

[Device Unknown]

I am in the process of writing an article that is along these lines. My questions to a lead AMD exec, seems to state that they are 110% intent on doing it right this next time. There won't be another "bulldozer fiasco" Their next new architecture will be fantastic.

 

[pwnorbpwnd]

Just because AMD is behind on performance doesn't mean they aren't getting competitive sales with the price to performance they have.

 

[jimmysmitty]

[citation][nom]fazers_on_stun[/nom]Well I guess this puts the AMD community reporter's statement that everything between AMD and GF is hunkey-dorey to rest..[/citation]

We called it, didn't we? I wonder what GF will do without AMD. They will have to either shape up or die out. But then who will FAB their 32nm parts.....

 

[pedro_mann]

[citation][nom]freggo[/nom]AMD will ALWAYS be behind; as Intel effectively controls the technology and the market. Naturally, AMD can just play 'follow the leader'. So if Intel comes out with something new, surprise... AMD will be behind and catching up; again. Frustrating position to be in.[/citation]
Historically this has not been true. Can we say AMD64 anyone? or perhaps integrated memory contollers, or the fact that Intels graphics are a joke? I would not agree that AMD is a follow the leader company, and they have at times and will in the future bring innnovative ideas to market. The one are they are always behind those is manufacturing process. And that is the one advantage that keeps intel in the lead.

 

[Guest]

This is a good move for AMD. GF sucks! TSMC is a good potential partner. This may be a set back as far as 2012 is concerned, but will be a step up for the future. While we keep talking Intel, AMD is worried from both sides of the fence. They have Intel to worry about as the elephant in the room, but they also have a fast growing market from beneath them called ARM that will be available in late 2012 as a PC platform. They needed to make a move and do it in a hurry. They can now position themselves to compete in 2013 when things are going to get interesting with ARM and a newly revamped ATOM on the low end, plus Intel Haswell on the top end. They could all put the squeeze on AMD and make them pretty non existent if they didn't make some serious changes. ARM is a bigger threat right now to AMD than it is to Intel because it will fill a value market segment.

 

[Guest]

Didn't Intel originally make an integrated controller for the P6 derivative, but scrapped that project and went to the PII and it's great Win95 performance (NT wasn't the big seller back then - circa '96). I do recall that Intel was developing an integrated memory controller and AMD certainly wasn't the first company to do that anyway. Noow x64 is definitely all AMD and they were the first to come to the market with 1 GHz and a 'true' multicore PC chip.