Amd news

Status
Fechado a novas mensagens.
AMD to launch Radeon-branded SSDs

Watch out world, AMD is going to launch solid state drives (SSDs) with their own branding soon.

It seems that AMD continues to bet on their AMD Radeon brand for expanding in to the consumer PC market. From Radeon-branded memory modules for desktop PCs and notebooks (from Patriot, VisionTek and XFX), to the AMD Radeon RAMDisk application from Dataram, AMD started expanding the Radeon brand into new component categories. From the looks of things, AMD isn’t satisfied with just two new products, and obviously didn’t want to slow down from reaching their VISION of making a PC system with all AMD branded components. To fix this, they added SSDs, but there are just a few questions that have yet to be answered.

First question: who will make them? AMD doesn’t have a memory division to support the R&D needed to make their own memory products and Spansion was spun off around 7 years ago. It’s easy to narrow the list down to a small number of players, but there are not enough hints to pick a specific company. Then there is the most important question, why bother to make your own?

As AMD is keen to position these AMD memory products to fill the missing parts in their consumer PC platforms, so finding ODM partners for off-the-shelf parts with their own branding would have shortest time-to-market with the lowest cost. So it’s a “pure marketing decision” that will hopefully bring in money. They have the products and they have the existing brand, so AMD will naturally use platform compatibility, stability and probably ‘platformance’ as selling points. AMD can also bundle several Radeon branded products to raise consumer awareness about these new products as well as the new-and-expanded AMD Radeon brand.

Marketing however is not traditionally a strength at AMD, especially since they laid off most of their product marketing people and outsourced to some PR firm who probably doesn’t have a clue, there are some worrying signs. The most prominent sign would be the “Good, Better, the Best” rating system from the AMD Memory module launch presentation about a year ago. It was aimed at “simplifying the purchase decision by not telling you any technical details”, a tactic that often annoys the targeted enthusiast market.

This rating system doesn’t give a clear sign of what to expect on the different price points, especially for memory and SSDs with a bunch of differentiating factors that are well hidden in the hard specifications. For SSDs, these factors would include types of NAND flash (SLC/MLC/eMLC), SSD capacity, SSD controller used, number of I/O channels, etc., none of which are addressed in AMDs rating system. The only upside of this system would be conveying to the customers that there is a range of choices, but most know that from the price. Over simplifying things doesn’t help anything, it just creates a lot more confusion and makes it hard to differentiate AMD products from other brands.

The next question is given the series of pure marketing decisions to enter the ever-changing and low margin memory and SSD markets, why bother in the first place? Although we don’t quite believe in the saying that AMD is about to enter system integration business, but they could if they wanted to. AMD is going to be making servers in a few years, are PCs and gadgets (read tablets and smartphones) going to follow? Will they be as stupid as Microsoft and turn the friendship of current OEM partners in to hostility? We really doubt they are this dumb, especially since many of the top AMD brass come from an OEM background.

It’s not impossible for AMD to continue this marketing-oriented path of decision making and attempt to mandate the use of Radeon branded products to OEMs and SIs to meet future AMD VISION sticker requirements. It is likely that they will provide reference designs to different levels in the “Good, Better, the Best” system in the future, and throw in some savings on memory and SSDs when building branded platforms. Sarcastically speaking, if this kind of plan is carried out successfully, then what we will be seeing is another Apple, but without the software, design, and company scale to make much of a market impact. No matter which way AMD goes, we feel that it’s not a good way to go.


Keep an eye out for these products as they will soon appear on the shelves of your local computer store or in any major e-tailers. The author sincerely hopes that AMD doesn’t create another round of confusion for customers, but the marketing strategy doesn’t encourage us much. <sarcasm> And who knows if AMD will launch AMD Radeon-branded PSUs or even AMD Radeon-branded chassis next time? Think of the fire red gradient on the Radeon brand, would look nice with black, and probably gold or silver trim too…
Clicar para expandir...

http://semiaccurate.com/2012/11/28/amd-to-launch-radeon-branded-ssds/
 
Será bem vindo, desde que tenha performance, preços e fiabilidade que venham eles :)

A AMD poderá construir os seus próprios SSD, com controlador próprio como a Samsung o faz.
 
Radeon SSDs... isto deve ser para small form factors. Deve baixar o custo de produção de um sistema desses para competir com os da Intel no mesmo ramo.
 
AMD Pushes Radeon HD 8000 "Sea Islands" Launch to 2Q-2013

According to industry sources, AMD pushed launch of its next-generation Radeon HD 8000 "Sea Islands" family of GPUs to the second quarter of 2013 (April-June). The delay comes in the wake of the company going through reorganization within its ranks, in response to poor market performance in Q3-2012. The company originally planned to launch its GPUs before the end of 2012. NVIDIA plans to launch a refreshed lineup of Kepler GPUs in 2013.



Source: DigiTimes
 
[h=2]Microsoft to Ditch NVIDIA and Intel with 2nd Gen. Surface[/h]
It looks like Microsoft wants to cast NVIDIA and Intel out of its second generation Surface plans, in a likely effort to cut cost of hardware (on its end anyway). On the receiving end of things at their expense are AMD and Qualcomm. Redmond also plans 14.6-inch Surface Book. Turning the rumor mill is a leak from @MS_nerd, a fairly reliable source of leaks from Redmond.
The second generation Surface line of products, slated for next year, will see a Qualcomm-made chipset (probably just Snapdragon quad-core), replace NVIDIA Tegra 3 in the Surface RT; and AMD's "Temash" line of low-power quad-core APUs replace Intel Core i5 ULV dual-core in Surface Pro.
The Surface RT 2.0 will be built into the 8.9-inch form-factor, breaking away from the current 10.6-inch common physical size between the Surface RT and Pro. The 8.9-inch display could end up with higher pixel density, if it retains its current 1366 x 768 pixels resolution. The Surface Pro, on the other hand, could feature a larger 11.6-inch display. Since it's already beaming out 1920 x 1080 pixels, we don't expect the resolution to go up. If anything, the move toward a larger display is to spice up its desktop/laptop credentials.
The last sack of flour from the rumor mill reads that Microsoft is working on an even larger 14.6-inch device called Surface Book. In all likelihood, this could be a hybrid-Ultrabook with a solid keyboard/trackpad. Fortunately for Intel, it retains the powerful Core i5 "Haswell" processor.
Clicar para expandir...

Se isto se confirma, serão grandes notícias para a AMD. O Jaguar deve ser promissor para estar a gerar este interesse.
 
Update ao roadmap dos servidores

2mqqjcp.jpg

http://www.cpu-world.com/news_2012/2012120301_AMD_server_CPU_roadmap_for_2013_and_2014.html

28nm apenas em 2014, o que indica que talvez o "leaked" roadmap divulgado pelo donanimhaber.com seja verdadeiro, e talvez esteja aqui a explicação do porquê do M.I.A. do core "Steamroller" e do "Kaveri" (atraso dos 28nm na GF?) mas nada diz acerca da base dos "future high power CPU", pelo que se aguardam os próximos capítulos da novela "Steamroller". Mas pelo menos a novela "AMD cancela os BIG core CPU" deverá estar nos últimos episódios.

Assim provavelmente será o "Excavator" a não ver a luz do dia, e nessa altura seja lançada uma nova arquitectura.
 
AMD Announces Opteron 4300 and 3300 Series Processors

AMD today unveiled nine new mid-range and entry-level AMD Opteron 4300 Series and 3300 Series server processors increasing performance per watt versus the previous generation to maximize compute capabilities in power-constrained environments1, while delivering outstanding value to cloud providers and web hosts. The AMD Opteron 3300 Series processors provide enterprise-class features in a low power envelope at a desktop processor price, making it ideal for web hosts and small- and medium-sized businesses.

The AMD Opteron 4300 Series processors offer up to 15 percent better performance versus the prior generation, based on SPECint benchmark results2. And SPECpower results show up to a 24 percent improvement in performance per watt. As with the AMD Opteron 6300 Series processors, these are socket compatible with the previous generation processor, providing an effortless upgrade path for customers.
"The Piledriver core architecture shared by the AMD Opteron 4300 and 3300 Series processors provides optimized performance, power and price for today's customer," said Suresh Gopalakrishnan, general manager, Server Business Unit, AMD. "These new processors are ideal for cloud providers, web hosts and small- and medium-sized businesses who want to address their space and power constraints. With a simple upgrade, existing customers can obtain more performance and improved energy efficiency and new customers will obtain compelling value and ROI."

Extensive Software Support
Deep and broad software ecosystem support is available for the new processors. Several of the key tools include:
  • Hypervisors: Citrix XenServer, Microsoft Windows Server 2012 Hyper-V, Red Hat Enterprise Virtualization, and VMware vSphere;
  • Operating systems: Microsoft Windows, SUSE Linux Enterprise and Red Hat Enterprise Linux;
  • Key developer tools: AMD Open64 compilers and AMD Core Math Library, GNU Compiler Collection (GCC), Microsoft Visual Studio 2012, and The Portland Group (PGI) compilers;
  • Java virtual machine: AMD Aparapi, OpenJDK Project "Sumatra," Oracle JVM and IBM JVM.
Partner Traction
Socket, thermal and chipset consistency with previous generation AMD Opteron 4200 and 3200 Series processors allows OEMs to easily carry forward existing server designs and enables end customers to deploy new server hardware within their existing power and cooling infrastructure. Servers based on the new processors are now available from Supermicro as well as the AMD SeaMicro SM15000 platform. Additional server offerings will soon be available from Dell and others.

"AMD Opteron processor solutions have provided excellent investment return for our customers," said Brian Payne, executive director of Server Solutions, Dell. "The new 4300 and 3300 Series processors in the AMD-based Dell PowerEdge servers maintain this history and are ideal for web service providers who strive to maximize revenue from their server infrastructure and also for small- and medium-businesses who carefully invest to enable growth."

"Supermicro has industry-leading server solutions and motherboards in production supporting AMD Opteron 4300 and 3300 Series processors," said Don Clegg, vice president of Marketing and Business Development at Supermicro. "Working closely with AMD, Supermicro is delivering 1U short-depth SuperServers using the AMD Opteron 4300 and 3300 Series processors, as well as a 3U 12-node MicroCloud optimized for the AMD Opteron 3300 Series processor for hosting and cloud providers. With systems using these new processors and the recently launched AMD Opteron 6300 Series CPU, Supermicro's time-to-market delivery provides customers a full range of Opteron-based solutions that cover extremely cost-sensitive web- and cloud-hosting to high-end HPC applications."

Key Facts, Performance and Technical Detail - AMD Opteron 4300 Series Processors
  • Frequency: Up to 3.4 GHz base frequency and up to 3.8 GHz using AMD Turbo CORE technology for performance boost when applications demand it;
  • Up to eight cores per socket for scaling on thread-intensive environments;
  • As little as 35W TDP for power-constrained environments;
  • AMD-P power management features maximize performance per-watt across workloads while giving IT managers more control over their data center power profiles;
  • The only x86 processor to support ultra-low voltage 1.25V memory;
  • AMD Virtualization (AMD-V) technology for near-native performance in virtualized environments and lowest cost per virtual machine (VM);
  • Two memory channels with up to 1866 MHz memory support;
  • Supports up to six DIMMs per CPU for up to 192GB memory per processor;
  • Up to two x16 HyperTransport technology (HT3) links at up to 6.4GT/s per link.
Key Facts, Performance and Technical Detail - AMD Opteron 3300 Series Processors
  • Frequency: Up to 2.8 GHz base frequency and up to 3.8 GHz using AMD Turbo CORE technology for performance boost when applications demand it;
  • Up to eight cores per socket for web hosting upsell opportunities;
  • As little as 25W TDP for power constrained environments;
  • AMD-P power management features maximize performance-per-watt across workloads while giving IT managers more control over their data center power profiles;
  • AMD-V technology for low cost virtualization;
  • Two memory channels with up to 1866 MHz memory support;
  • Supports up to four DIMMs per CPU for up to 32GB memory per processor;
  • One x16 HyperTransport technology (HT3) link at up to 5.2GT/s.
For more information, refer to the data sheets of Opteron 4300 series and Opteron 3300 series.
Clicar para expandir...

http://www.techpowerup.com/176571/AMD-Announces-Opteron-4300-and-3300-Series-Processors.html


 
Última edição:
AMD: We're not abandoning socketed CPUs

AMD has a long history of supporting the DIY and enthusiast desktop market with socketed CPUs & APUs that are compatible with a wide range of motherboard products from our partners. That will continue through 2013 and 2014 with the "Kaveri" APU and FX CPU lines. We have no plans at this time to move to BGA only packaging and look forward to continuing to support this critical segment of the market.
As the company that introduced new types of BGA packages in ultrathin platforms several years ago, and today offers BGA-packaged processors for everything from ultrathin notebooks to all-in-one desktops, to embedded applications and tablets, we certainly understand Intel's enthusiasm for the approach. But for the desktop market, and the enthusiasts with whom AMD has built its brand, we understand what matters to them and how we can continue to bring better value and a better experience.
Clicar para expandir...

Pelo menos a AMD não planeia abandonar os seus clientes com gosto pelo hardware.
 
lol seria de admirar se não tivessem de contrariar uma notícia como essa dos CPU soldados. Há que angariar clientes com boa publicidade. Agora resta saber se vale a pena ter um CPU não soldado. Vamos ver se a performance não compensa mais ter um soldado.
 
SubCode disse:
lol seria de admirar se não tivessem de contrariar uma notícia como essa dos CPU soldados. Há que angariar clientes com boa publicidade. Agora resta saber se vale a pena ter um CPU não soldado. Vamos ver se a performance não compensa mais ter um soldado.
Clicar para expandir...

nao se trata de performance, mas sim os custos que terás se um dos componentes queimar.
 
As coisas continuam a correr muito mal para a AMD. Agora vai ter de pagar 320 milhões à GF pelo cancelamento de encomendas de waffers.

AMD reduces GlobalFoundries wafer order, penalized $320 million

AMD’s prospects just keep taking a turn for the worse as we approach 2013. The chip maker has suffered from a slow CPU market and increased pressure from rival Intel which has subsequently forced them to reduce their wafer purchase commitments with GlobalFoundries.

The decision to break contract with GlobalFoundries was likely done to avoid having excess inventory that would result in further write-downs in the near future. AMD originally agreed to buy $500 million worth of silicon during the fourth quarter but now will only spend $115 million.

Of course, the semiconductor foundry didn’t just let AMD out of the contract without penalty. The chip maker agreed to a termination payment of $320 million that will be paid over several quarters as follows.

$80 million will be due by December 28, 2012, another $40 million will have to be paid by April 1, 2013 and AMD signed a promissory note valued at $200 million that’s due by December 31, 2013.

AMD said the termination fee will result in a one-time charge of $165 million that will show up on their fourth quarter results. In a press release on the matter, the company said they expect to return to free cash flow generation in the second half of 2013.

AMD CEO Rory Read said the announcement demonstrates that the long-term strategic partnership between the two companies continues to be beneficial on both sides. He praised GlobalFoundries for meeting their delivery requirement in 2012 and looks forward to developing and growing AMD with their help.
Clicar para expandir...
 
O problema não é só a diminuição das vendas, que naturalmente levou a uma redução das encomendas, a questão já vem desde o spin-off da GF e envolve as duas partes, a GF também já teve de suportar os custos quando "yields" dos Llano eram baixos, a AMD pagava por chips funcionais e não por wafers. O facto de o fabrico ser a 32nm SOI, que foi desenvolvido para a AMD (e do qual a AMD é a única cliente), implicava a aquisição de um determinado nº de wafers por parte da AMD, e mesmo assim não vão pagar a diferença dos 500 - 115.

Mas no meio dessa até acaba por vir uma que pode ser positiva:

those negotiations appear to have concluded, with the following results:

- Lowered wafer purchase commitments for 4Q12, to $115M;
- A $320M payment (spread out in three installments) to terminate the current take-or-pay agreement, associated with that adjusted 4Q12 commitment;
- Wafer purchase commitments in fiscal 2013 of $1.5B;
- Wafer purchase commitment of $250M in fiscal 1Q14;
- Reduction in further reimbursements to GlobalFoundries for R&D costs, as AMD moves to standard 28nm process technology.
Clicar para expandir...
http://www.electroiq.com/articles/sst/2012/12/amd-globalfoundries-new-wafer-deal-what-it-means.html


AMD will depart from SOI technology at the 28nm node and going forward will employ bulk technology instead. The new information in this regard is that this reduces reimbursements to GLOBALFOUNDRIES for future research and development costs. The savings are expected to be around $20 million per quarter which would help the company reach their new quarterly operating expense target of $450 million.

Unlike the limited waiver of exclusivity in their 2012 WSA, there is no such clause in the 2013 WSA.

"We are obligated to make all microprocessor products at GLOBALFOUNDRIES, and we will do that." said Devinder Kumar during the call.

What this means is that at 28nm and below, any CPU or APU will be made exclusively at GLOBALFOUNDRIES for the time being.
Clicar para expandir...
http://www.brightsideofnews.com/new...gets2c-revises-globalfoundries-agreement.aspx

O facto de não terem que pagar por um processo de fabrico "exclusivo" permite poupar nos custos de R&D, falta saber é se terá impacto a nível de consumo e/ou performance, não sei a diferença que fará no preço das wafers em si.
Mas hoje foram divulgadas mais umas declarações do CEO, Rory Read:

n a bid to cut costs, Advanced Micro Devices claims it is turning down certain low-volume deals that require it to invest into implementation of its products.

“Not all design wins are created equal. Actually, more is not better. More equals more complexity, it means testing platforms, back-end costs that drive [company’s] costs. […] What I am going to do in our space is to focus on design wins that are going to drive volume. […] Every design has associated cost with testing, putting on the platform, etc.,” said Rory Read, chief executive of Advanced Mice Devices, at Credit Suisse technology conference in late November.

As the time has shown, Nvidia is now the No. 1 supplier of notebook GPUs (based on data from Mercury Research provided by Nvidia) because of AMD’s reluctance to help integrate its Radeon Mobility products based on the recent architecture.

Mr. Read also complained about the production technologies used by AMD. Given the vast product lines, AMD used up to nine different process technologies at two of its foundry partners (45nm and 32nm at Globalfoundries, 55nm, 40nm, 28nm with or without LP features). Mr. Read wants AMD to only use two process technologies. Moreover, he wants to reduce the number of metal layers that contain logic elements within microprocessors to “industry standard” levels, which naturally means compromises in performance.
Clicar para expandir...
http://www.xbitlabs.com/news/cpu/di...e_Not_Interested_in_Low_Volume_Customers.html
 
Ai estão 3 pontos muito importantes para melhorar, tirar as limitações dos núcleos, melhorar o IPC e baixar os consumos.
Vamos lá ver se a AMD finalmente consegue tornar a arquitetura bulldozer em algo que apresente concorrência para a Intel.
 
Tenho uma dúvida esquesita. O piledriver em APU consegue chegar aos 4 GHz e ainda manter-se dentro dos 65W. Então porque é que não deitam fora a cache de L3 nos CPUs?

Ou perguntando de outra maneira... porque é que não fazem 'Athlons de 8 cores' sem L3 cache no FM2?
 
Última edição:
Porque nesse caso irias alimentar o CPU com menos dados. Sem a cache L3 o CPU teria de esperar por mais dados vindos da RAm dos sistema, que por norma é muito mais lenta do que a cache do CPU.
No caso do Bulldozer já tens um CPU que passa fome de dados, sem a cache L3 iria passar ainda mais, o que é o oposto do que a AMD quer com o Steamroller.
Aliás, um dos grandes defeitos do Bulldozer é ter uma cache lenta, em comparação aos Phenom II e ainda mais comparada com os Sandy Bridge.

41731.png
 
Status
Fechado a novas mensagens.
Back
Topo