[·feist·]

Sometimes you really do have to pay close attention to presentation slides.


Lots of info incoming.

AMD 2012 Financial Analyst Day
http://ir.amd.com/phoenix.zhtml?c=74...2012analystday <--- multiple presentation .PDFs available

Your new AMD decoder key
http://blogs.amd.com/work/2012/02/02...d-decoder-key/ <--- codenames no longer in use/canceled products

Compal Trinity ODM reference design eyes-on
http://www.engadget.com/2012/02/02/c...esign-eyes-on/

Not just a FAD, AMD aims at the market ultra-thin laptop-like device
http://www.pcper.com/news/General-Te...-laptop-device

AMD Trinity ULV 3DMark scores revealed
http://vr-zone.com/articles/amd-trin...led/14742.html

Inside this particular prototype is one of the lower voltage variants of Trinity (read: either the 17W or 25W part), which enables that svelte 18mm profile.

An estimated starting price of half an Ultrabook (roughly in the $500 to $600 range).

There weren't many details though Su did state they were hoping for prices in the $600-800 range could put a lot of pressure on Intel.

One of the benefits Trinity will bring is what AMD called 'All day' battery life, with a 12 hour lifespan predicted. Trinity uses half the power of Llano as well as featuring an improved graphics core which they predict to be half again as powerful as Intel's HD Graphics. They also predict the new Bulldozer architecture will increase general computing power.

Compal, a contract maker of electronics, may start producing notebooks using this design for interested parties, whereas other notebook makers may create their own ultra-thin mobile computers powered by AMD Trinity. "The 18mm reference design from Compal is what many OEMs are looking at."

To counter Intel's ultrabook initiative, AMD is preparing fully enabled quad-core (two module) APUs for ultrathins. The typical AMD ultrathin will be around ~18mm, featuring 17W APUs, while slightly thicker ultrathins may use high performance 25W APUs.

AMD revealed typical 3D Mark Vantage scores for selected 17W and 25W SKUs, performed on Compal reference designs. A A6 ULV 17W Trinity scores 2355 3D Marks. This is twice as fast as Sandy Bridge Core i5 2537M, an ultrabook regular, which scores 1158 3D Marks. Of course, Trinity will be competing with Ivy Bridge ultrabooks, but AMD expects the given A6 ULV to still be over 50% faster than competing Ivy Bridge ultrabooks. The pinnacle of performance/power lies with the A10 Low Voltage APU, however. The 25W LV APU scores a blistering 3600 3D Marks, which sets a new record for integrated graphics in a laptop and puts most entry-level mobile discrete GPUs to shame. Trinity will also be present in performance laptops in 35W and 45W variants, which will further advance integrated graphics in notebooks. On the CPU front, Ivy Bridge should be dominant in less threaded benchmarks. However, things will be much closer in multi-threaded benchmarks as AMD is offering two-module / quad-core APUs against Intel's dual-core / four hyper-thread CPUs. With the right price points, AMD is well positioned to offer a compelling alternative to Intel's pricey ultrabooks.

AMD Unwraps 2013 Client Roadmap: 28nm Everywhere.
http://www.xbitlabs.com/news/cpu/dis...verywhere.html

AMD's 2012 - 2013 Client CPU/GPU/APU Roadmap Revealed
http://www.anandtech.com/show/5491/a...admap-revealed

All product lineups that AMD promises to release next year will be substantially different from what is available today, hence, they will not be just product shrinks or refreshes. 28nm process technology alone is likely to allow AMD to boost performance and feature-set of its chips without increasing power consumption compared to existing solutions that are made using 32nm or 40nm fabrication processes.

The new Steamroller x86 high-performance cores will improve performance by boosting efficiency of execution and will also include design elements from low-power architectures to reduce power consumption.

Going into 2013 AMD will move all mainstream client APUs to 28nm and bring a GCN (Graphics Core Next) based GPU to all of the APUs. Kaveri, the Llano/Trinity follow-on, will use Steamroller cores (evolution of Bulldozer/Piledriver) while Kabini and Temash will use Jaguar. Jaguar is an evolution of the Bobcat core although we don't have any architectural details at this time. Kabini and Temash will also integrate the Fusion Controller Hub (FCH, aka South Bridge) making these two APUs AMD's first true single-chip solutions. AMD's FX platform will get an update to Piledriver cores this year with Vishera. There's no visibility beyond Vishera unfortunately, although it's probably a safe bet that we'll see a Steamroller based derivative at some point. AMD's 2013 roadmap is heavily built around HSA. The hope is that with Graphics Core Next on-die, and proper software support, AMD will be able to deliver a compelling heterogenous computing platform that lets you leverage the strengths of both x86 CPU cores and a GPU built for compute. AMD has been chasing the promise of heterogenous compute for a while now, but its roadmap is clearly built around that vision becoming a reality.

Following the cancelation of Wichita and Krishna, Brazos 2.0 will have to hold ground in the low-power market covering C-series and A-series for 2012. Brazos 2.0 is reported to be a minor improvement over Brazos. The main improvements are in the chipset as well as the introduction of Turbo Core. Hondo is the successor to Z-01 (Desna) and will appear in tablets and other ultra low power devices. Like Brazos 2.0, Hondo's improvements are also mostly in the FCH - power optimizations will cut down consumption significantly over Z-01. Hondo is rumoured to be branded Z-03. Both Hondo and Brazos 2.0 are based on the same infrastructure and fabricated by TSMC's 40nm process.

AMD sets out its plans for 2013, hints at a possible ARM future
http://arstechnica.com/business/news...arm-future.ars

AMD Promises "Full" Fusion of CPU and GPU in 2014
http://www.xbitlabs.com/news/cpu/dis...U_in_2014.html

Understanding AMD's Roadmap & New Direction
http://www.anandtech.com/show/5503/u...-new-direction

The underlying themes to AMD's plans are faster iteration—a GPU-like 18-24 months between CPU designs, compared to the current 3 or more years—achieved by moving away from custom designs and depending more heavily on synthesized chip layouts, and lower power usage. This in turn will give AMD more flexibility to integrate CPUs and GPUs—and potentially other co-processors too—into what the company calls APUs (accelerated processing units).

Now it's official AMD policy to have shorter design cycles. These shorter design cycles will leverage lower amounts of custom block design and lean more on easily synthesizable architectures. The tradeoff is obviously performance but you do get better time to market. As was the case with Brazos however, if you can bring the right combination of technologies to market at the right time, the tradeoff is worth it.

AMD's goal with HSA is to make mixed workloads that use both the CPU and the GPU easier. The 2013 HSA (Heterogeneous Systems Architecture) processors will give the GPU and CPU a unified, coherent address space, with the GPU able to use the same demand-paged virtual memory that the CPU uses. This means that data will no longer have to be moved from CPU to GPU to allow the GPU to work on it, and that both processors will be able to operate on the same data simultaneously, making mixed CPU/GPU computation seamless.

This year we got Graphics Core Next, but next year we'll see a unified address space that both AMD CPUs and GPUs can access (today CPUs and GPUs mostly store separate copies of data in separate memory spaces). In 2014 AMD plans to deliver HSA compatible GPUs that allow for true heterogeneous computing where workloads will run, seamlessly, on both CPUs and GPUs in parallel. The latter is something we've been waiting on for years now but AMD seems committed to delivering it in a major way in just two years.

Given the limitations of the current architecture, x86 and Radeon cores should use dedicated memory, which tends to be inefficient. Essentially, nowadays the software completely controls which compute resources to use and when. But after years of evolution, which will involve development of both hardware and software/compilers/tools, accelerated processing units in 2014 will be able to dynamically decide (possibly, when it comes to new programs) which task is better to execute on a particular core thanks to new software as well as special features of the chips.

Dynamic context switching between different types of cores will not only greatly speed up performance of such chips, but will also optimize power consumption as the most efficient hardware will be used to perform an operation. Earlier AMD expected to release "fully fused" Fusion chips in 2015 or beyond.

A mixed ISA future? With APUs, more easily iterated synthesized designs, and HSA, AMD is taking some big steps toward producing flexible, heterogeneous processors: processors that pack together many different cores, each with their own strengths and weaknesses. AMD wants to extend the APU concept with other processor units, both AMD-originated and third party, to offer customers tailored solutions suitable for different applications. For example, motion video codec accelerators (such as those found in Intel's Medfield system-on-chip) would be attractive in tablets and, if AMD can get power usage down enough, even smartphones (though this is not a market the company is targeting at present). One particularly intriguing third-party unit would be an ARM processor. AMD mentioned ARM several times during its presentations, and a number of its slides stated that the company wanted to produce SoCs that are "ambidextrous... across ISAs," stating also that the company was "flexible around ISA." AMD spoke of these mixed-ISA processors in the context of servers and datacenters, so the immediate utility of an ARM processor is not clear. However, ARM Ltd wants to move ARM into the server space, having recently extended the ISA to support 64-bit systems. If ARM were to become a significant force in this market, the ability to natively run both ARM and x86 workloads on a single chip might become attractive.

Looking at TI, Qualcomm, NVIDIA and even Intel, integrating 3rd party IP into an SoC isn't unusual - particularly when competing in the ultra mobile space. AMD wants the same flexibility. Going forward, if AMD is successful, we will see SoCs based on AMD technologies that are combined with 3rd party IP. In theory this could come in the form of anything from a video decoder/encoder block to an ARM based CPU/GPU.

AMD is mostly interested in markets that have high annual growth rates. Looking [below] you can see that pretty much all of those categories with the exception of the client desktop are interesting for AMD. It's about time that AMD focused more on mobile and I don't believe that it's too late for the company.

From a product standpoint, AMD is really focusing on its mainstream and entry level APUs. Rory didn't come out and say it here but no where in AMD's future direction is a focus on the high-end x86 CPU space. Also note that AMD isn't going to be as focused on delivering high performance products on the absolute latest process node. It views Brazos as one of its biggest successes to date and that architecture was built on a 40nm process with an easily synthesizable architecture. It's likely that the future of AMD is built around more of these easy to manufacture SoCs rather than highly custom, bleeding edge CPUs. AMD plans on leveraging OEMs to deliver its products but it also wants to explore other routes as well. Rory referenced the game console model, where AMD would sell an ODM a chip solution tailored specifically to their needs. AMD wants to use this model to complement the more traditional route of selling its products. The transition here makes sense if you look at the current tablet space. The SoC players in tablets effectively follow the game console model. You buy a tablet that has an SoC that's custom tailored to its needs rather than buying a system with a myriad of CPU options.

Originally AMD had talked about introducing a new G2012 platform and delivering 10 & 20-core solutions called Sepang and Terramar. Those plans have been scrapped for the moment and what we get instead is a drop-in replacement for existing Opteron 6200 CPUs. Take the current 6200 lineup, upgrade the CPU cores to Piledriver and you get a high level look at AMD's near-term server strategy. The sockets remain the same, as do the core counts, but performance should go up. AMD hasn't given us any more detail as to what Piledriver fixes other than to say that it's a higher IPC version of Bulldozer.

Related:

Solid Third Quarter Has Intel So Giddy, It Praises AMD
http://www.pcmag.com/article2/0,2817,2394923,00.asp

"I hear this argument a lot about ARM coming into the PC segment of the market," Smith said. "That they're going to come into the low-end of the market with a $30 processor to compete with our $100 processor and then they win. "But we have $30 processors, full kits in fact, that we make a nice profit on. And AMD has a $20 kit."

Still, it seems like the last time Intel execs discussed the company's long-time archenemy this much was back when Intel was handing over $1.25 billion to AMD to settle a bunch of anti-trust lawsuits. The last time Intel talked up AMD this much in a non-legal context was probably back when AMD's K7 architecture was kicking butt and taking names.

The death of CPU scaling: From one core to many — and why we’re still stuck
http://www.extremetech.com/computing...re-still-stuck

01 - Intro 02 - The multi-core swerve 03 - The rise (and limit) of Many-Core