RSSAMD-ATI to Offer CPU/GPU Combo
By Scott M. Fulton, III | Published October 25, 2006, 11:34 AM
If there was any doubt that CPU maker AMD’s principal reason for acquiring graphics chip powerhouse ATI was to build a mobile computing platform that would rival Intel’s Centrino, it was erased this morning with AMD’s announcement of a platform project it’s currently calling “Fusion.”
Ostensibly, the purpose of this morning’s AMD statement was to announce that it had completed its merger with ATI. As a visit to ATI's Web site will also make clear, AMD is clearly in the driver’s seat, although the ATI brand will apparently continue to adorn Radeon graphics cards for the near future.
But this morning’s comment from AMD’s CTO, Phil Hester, fires a broadside squarely at Intel’s design philosophy of multicore schemes in multiples of two:
“In this increasingly diverse x86 computing environment, simply adding more CPU cores to a baseline architecture will not be enough,” said Hester. “As x86 scales from palmtops to petaFLOPS, modular processor designs leveraging both CPU and GPU compute capabilities will be essential in meeting the requirements of computing in 2008 and beyond.”
In other words, AMD’s future “Fusion” platform, which the company says could be ready by late 2008 or early 2009, won’t just be an Athlon and a Radeon sharing the same silicon. The companies will truly be looking into the potential benefits of using multiple pipelines along with multiple cores, as a new approach to parallelism in computing tasks.
When the two companies first announced their merger intentions last July, although both sides were using strange new hybrid almost-terms like “GP-CPU” to describe the research they had read about –- if not actually participated in –- their willingness to head down this new architectural road seemed more tentative than definitive.
With three months having passed, and analysts throughout the duration of that time having discussed the real possibility of Intel’s having recaptured the market share momentum, the new AMD seems more willing to embrace a development path that could more clearly distinguish itself from its core competitor.
Almost from the day that pipeline processing was added to the first hardware-assisted rendering systems for PCs, researchers (though mostly with universities, not manufacturers) have been exploring the notion of co-opting graphics processor-style parallelism for use with everyday tasks. Both multicore CPUs and modern GPUs use parallelism to divide and conquer complex tasks, although their approaches are radically different from one another. That difference, researchers say, could make them complementary rather than contradictory.
Multicore parallelism delegates discrete computing tasks among differentiated logic units. If you can imagine the CPU performing triage when it first examines a passage of object code, it delegates tasks into separate operating rooms where individual doctors, if you will, can concentrate upon each one exclusively.
The approach the CPU takes depends on both its own architecture and that of the program. In discrete parallelism, programs are compiled by software that already “knows” how tasks should be delegated, so the compilers “toe-tag” tasks in advance. The processor reads these tags and assigns the appropriate logic units, without much decision involved on the CPU’s part.
This is the kind of “discrete parallelism” approach that Intel’s Itanium and Itanium-2 series (IA-32 and IA-64 architectures, respectively) take. As you probably already know, the fact that Itaniums have not been hardware-compatible with x86 architecture, has been one of Intel’s historical detriments.
The other approach CPUs take is with implied parallelism where the processor analyzes a program that would otherwise run perfectly well in a single-core system, but divides the stream into tasks as it sees fit. This is how Intel’s hyperthreading architecture worked during its short-lived, pre-multicore era; and it’s how x86 programs can run in multicore x86 systems, running Core 2 Duo or Athlon processors, today.
But there are techniques programmers could take to help out the implied parallelism process, usually involving compiler switches, including on Intel’s own compilers. It wouldn’t change the program much, except for possibly a few extra pragmas and some tighter restrictions on types. Still, as Intel’s own developers have noted, programmers aren’t often willing to make these quick, subtle changes to their code, even when the opportunity is right in front of them.
Keep that in mind for a moment as we compare these to GPU-style parallelism. In a GPU, there aren’t multiple cores or logic units. Instead, since their key purpose is often to render vast scenes using the same math over and over, often literally millions of times repeated, GPUs parallelize their data structures. They don’t delegate multiple tasks; they simply make the same repetitive task more efficient by replicating it over multiple pipes simultaneously. It’s as though the same doctor could perform the same operation through a monitor, on multiple patients with the same affliction simultaneously.
Although CPU manufacturers arguably pioneered the concept of Single Instruction/Multiple Data for use with the first graphics acceleration drivers in the late 1980s, it’s GPU manufacturers such as ATI and Nvidia which took that ball and ran with it, to some extent clear out of the CPU companies’ ballpark.
As university researchers have learned, co-opting the GPU for use in processing everyday, complex data structures, yields unprecedented processing speed, on a scale considered commensurate with supercomputers just a few years ago.
But unlike the case with implied parallelism in multicore CPUs, a shift to SIMD-style programming architecture would, by most analyses, mean a monumental transformation in how programs are written. We’re not talking about compiler switches any more.
Several new questions and concerns emerge: How does the new AMD evangelize programmers into making the changes necessary to embrace this new platform, in only two years’ time, while simultaneously (to borrow a phrase from parallelism) trumpeting “Fusion’s” compatibility with x86 architecture. When Itanium weighted down Intel’s momentum, AMD didn’t pull any punches in pointing that out to everyone. It’s made astonishing success by touting its fully-compatible, x86-oriented approach to 64-bit computing, at Intel’s expense.
AMD cannot afford to make a U-turn in its philosophy now, not with Intel looking for every weakness it could possibly exploit. And if AMD succeeds in hybridizing an x86/Radeon platform while staying compatible and holding true to its word, if there are no automatic performance gains in the process of upgrading –- as there clearly are with multi-core -– will customers see the effort as a benefit to them?
Paralell processing is the key ,therefore programmers must be the first to embrace this and I believe that they will.Paralellism is not a small niche of the market anymore it is growing everyday and lack of eficent software is the only holdback.
Quad core is only the start look at Suns Niagara CPU ,basically a large group of simple CPU cores all on one die.
Take Niagara a step further and integrate graphics into each core and the benefits are amazing ,they just scream "Notebook".
Imagine faster processing ,lower power consumption ,fewer parts ,lower production cost.
ATI was bought for its IP portfolio and to make chipsets ,not to kill the ATI brand.
Score: 0
Hopefully whatever product that will come out from this integration in the form of a unified platform of CPU-GPU will significantly lower the cost for us end-users considering the ever increasing cost of owning either discrete CPU and GPU components nowadays.
Btw, first post.
Score: 0
I'm interested to see where this goes... Kinda early in the game to make any clear cut decision.
Score: 0
apparently Intel was working on something like that and then dropped it. I like th idea of another socket on the board for the gfx card. imagine plugging your gpu, and the video ram right into the mamaboard. insane
Score: 0
I'm really interested to see how the merger will effect ATI Catalyst driver support. Hopefully for the better!
Score: 0
doesn't lookit! :p
I think ati owners (me presently) have to wait longer than usual now. Too long to begin with! And good drivers mind you, lol.
next stop...nvidia
Score: 0
Why does it sound like they are trying to do an Amiga type architecture on a chip.
Score: 0
The Amiga had dedicated chips. Denise did the Graphics, Paula did the sound and Agnus did the rest.
So long as it can be bypassed via a PCIe slot I dont really see a problem, Maybe even designed to offload tasks to the internal one can make it more powerfull. If its designed well it could be better for the user than to just make cheaper higher spec machines.
http://en.wikipedia.org/.../Original_Amiga_chipset
Score: 0
Don't forget Gary, he controlled the flow..
;-)
Score: 0
I most of the commenters here are missing the point. It's not about power consumption or beating Intel in having this combo.
The big picture here is being able to use the GPU to process instructions at the same time as the normal CPU. If you look at how fast the GPUs are today, compared to how fast the CPUs are. If you are looking to do some serious realtime processing (medical research, simulations (like - how well will this new plane design work or what will the weather be like tomorrow) - the GPUs can give the system an tremendous boost.
Maybe I'm just old, but I remember back when there was no FPU (floating point unit) built into the CPU. If you wanted to do a lot of calculations (in spreadsheets - not games) you had to wait on the CPU. Intel had a separate FPU that you could install right next door to the CPU to do the job better.
That's kind of what this is all about. Except now we're going to put a honking fast FPU right in the processor.
I think that you're still going to have a separate graphics card, but this push in speed in calculating will really give AMD an even bigger edge of Intel.
Right now the high end graphics cards can display a moving object rotating through space. But the CPU has to figure out how to rotate it - that's the bottle neck. Now we move the GPU calculation engine into the CPU and it will do it even faster and the GPU can spend even more time in showing a better picture.
Score: 0
What you say just doesn't make sense. There are DSPs way faster than any GPU on the market. The GPU technology is derived from DSP architectures of the past just like current CRISC x86 CPUs got not just MMX/SSE/3DNow! sub-units from DSPs but many of the improvements added during the years as well.
There is nothing really new about all this, but there can be good and bad engineering at improving and using DSP derived architectures. And creating a GP-CPU it's the dumbest idea ever. It would make sense for very low power approaches and markets like set top boxes, media players, HTPCs and so on but not with current and future CPUs and GPUs.
Score: 0
You have a very good point to make. But do you really want chip companies to dictate what onboard video you should use? I know that Dell loves this idea because they can lower their prices below eMachines standard.
Intel has been trying for years to promote onboard video, with very little success. I did put the Intel stuff to the test, and the results looked like a 1930s movie.
Now AMD want’s to shove ATI down our throats, because they assume that consumers are stupid and we need daddy Ruiz to show us the way Taco Bell.
The problem with video integration is that innovation from video card manufacturers has to slow down in order to keep up with chip companies. You know very well that when you buy a video card in a retail box, the drivers inside are already several versions old. If AMD want’s full video integration into their chips, then don’t expect much ATI innovation.
NVidia, 3Dlabs and Matrox are the only players left.
Score: 0
I think alot of people are jumping to the video replacement conclusion when that is not really what is going on here, the phrase in the article that stuck out to me was...
...graphics processor-style parallelism for use with everyday tasks
We are starting to hear about some developers trying to use advanced graphics features (shaders if I recall correctly) for physics calculations and other complex not directly graphics related tasks.
I think this article is insinuating MMX on strong steroids, not onboard video.
... or a heterogeneous multi-core architecture instead of the mass homogeneity Intel is heading towards.
Score: 0
If AMD is going to counter the threat from intel, it has to do things right.
With the best of AMD and ATI on the same chip, performance can be greatly improved. With not having to come off chip or though the bus, it could significantly alter the future of gaming. But im sure they will continue to do CPU only models.
Score: 0
Isn't this the same thing that the Cyrix MediaGX CPU did years ago, a CPU/GPU on one chip? Sounds almost like AMD is trying to claim that they are the first company to do this.
Score: 0
They don't need to create one, they already have a CPU with onboard GPU - the Geode line of CPUs.
IIRC.
Score: 0
Cyrix sold off to National Semiconductor. AMD acquired the patents to the MediaGX when they purchased the Geode processor. The What was left over from cyrix was sold to VIA.
So yes, in a way they are still the first because they purchased it.
Score: 0
The way I see it, things are going to change. You’ll start seeing AMD saying things like this video card is “optimized” to work with this CPU/system. And to me it makes sense to do stuff like that, but they’re not going to shoot themselves in the foot and share certain things with Nvidia. This will leave Nvidia out when it comes to the AMD platform, which will leave them with Intel. In the end Intel will be in bed with Nvidia. Wish it were the other way, but that’s life.
Score: 0
It's the dumbest idea ever, even worse than the failed Tejas Intel CPU project.
And unfortunately it seems that Intel is looking into it as well to not let AMD be the only one on the market with such a lame product.
Upcoming DirectX 10 GPUs like the Nvidia G80 are known to need an additional PSU and consuming over 150Watts ... so, what is a GP-CPU multi-core multi-shaders multi-pipeline ? 250-300Watts ?
That's just crazy. Power consumption has to go down a lot or it's just very bad engineering and they surely can't create a GP-CPU combo in 2-3 years from now able to consume less than a current GPU+CPU combo.
Intel CPUs after P4 Northwood core were a shame with Prescott and the failed never released Tejas core that reached 150Watts per core. And now AMD wants to engineer such a thing ?
Also, such an approach is going to destroy modularity and freedom to change the GPU at will on any PC. Changing a GPU with a new one will mean trashing the CPU as well, which means higher costs and compatibility issues than current modular approach. PCs are all about modularity, killing modularity kills the reason to buy a PC and threatens the IT market.
A GP-CPU could be used in super-computer clusters maybe where it's not needed to upgrade them but just adding more.
Score: 0
AMD has promised over and over (and they've stuck to it so far) to keep the wattage required by the CPU at 95. Last I heard from AMD, 4x4 was going to be no different, and I can't really see them being able to market this without sticking to their guns on that for this either.
...but we can comment on how bad or good it will be until pigs fly. Doesn't mean s*** until they actually release something.
Score: 0
>> "Also, such an approach is going to destroy modularity and freedom to change the GPU at will on any PC. Changing a GPU with a new one will mean trashing the CPU as well, which means higher costs and compatibility issues than current modular approach."
You clearly do not know what you are talking about in this comment. AMD is not going to force you to only use their GPU. There will be an option to use any video card that you want (as long as the chipset on the motherboard supports it).
Recall, AMD is all about choice and they are not going to force you to use an ATI GPU if you are set on nVidia.
Plus, AMD did not divorce nVidia at all. True, the relationship may be a little more bumpy, but AMD still proudly supports nVidia.
Score: 0
If their insane approach takes over the market no one will want to use video cards on bus slots because the CPUGPU bandwidth is obviously lower thru any kind of external bus.
Making a GP-CPU it's just insane marketing hype producing an insanely engineered project that just doesn't make sense. Also, GP-CPU will surely cost more than current CPUs to make and it will take many years for that extra cost to get absorbed and prices getting lower.
Score: 0
I think all of you who commented are missing the point. The main reason why AMD and ATI merged was to comete with Intel, not nVidia.
nVidia will not sway one way or the other, and if Intel does try to merge with them, they're going to pay a hefty cost for the merger (way more than AMD paid to merge with ATI).
nVidia has been making graphics chipsets for both platforms for a while. Why would they switch to Intel only to compete with ATI and drastically reduce their market?
I don't think it's a good idea for ATI (AMD) to limit the graphics chipset market to strictly AMD based computers, but I understand what they are trying to do. Intel already makes graphics chipsets alongside their processors, and this is more focused on lower end systems and laptops. This is where the AMD/ATI merger will emerge, and will offer direct competition with Intel's graphics/processor combination.
nVidia, on the other hand, will more than likely keep the same marketing strategy it has today, because nVidia has made a name for itself in both graphics and motherboard chipsets, whereas ATI, although for a few years has been making motherboard chipsets, is mainly known for graphics chipsets.
Both ATI and nVidia will continue to compete with each other on the medium and high end GPUs, whereas Intel has no share of that market.
Score: 0
i don't like this idea of GPU makers merging with CPU companies just like domio said that ati may have stop support for intel cheipsets. if that did or might happen that is scary. much to the fact that if nvidia and Intel merge too you not going to have much choice when it comes to what graphics you want to get with your CPU.
Score: 0
The combo sounds interesting. But looks can be deceiving.
Let’s go back a few months when ATI (just after they were bought by AMD) announced that certain drivers will no longer be updated for Intel chips. At the same time, future releases will only be tested with AMD chips. Shouldn't a video card manufacturer make chipsets for both chip companies and ensure that drivers always work? Is ATI becoming a biased company? Well, since they are now part of AMD, AMD will do anything to mess up with Intel.
So, don’t expect much from ATI anymore. They were a pretty good company. Now, the only independent and non-biased player is nVidia.
As for AMD, I don’t trust anything they say for the internet. I’m tired of all their PR venting rubbish. So, let’s see if they can deliver what they say. Otherwise it’s all speculation and probably an illusion to bump up the stock price.
Score: 0
In many ways even Nvidia cannot be un-biased anymore, as ATI joined with AMD, it will be only logical that Nvidia will partner with intel. Now, I'm far from suggesting Intel and Nvidia will merge--although I thought ATI and AMD wouldn't ever merge either, so anything's game--but at the least Nvidia and Intel will get closer over this for sure. After all, Intel needs Nvidia in some ways, while Nvidia needs Intel to launch most of its platforms. Again, I doubt nvidia will stop making chipsets for AMD altogether, but they will certainly be much more focused on Intel for now.
Score: 0
Intel pulled the bus license from ATI after they got word of the AMD+ATI thing--so ATI didn't really have a choice in the matter.
Read this article about the matter: http://www.theinquirer.n...ault.aspx?article=33225
You can't support future Intel platforms if Intel says, "Take a hike!"
Score: 0
I read that “very short” article that you mentioned. Unfortunately it doesn’t say anything as to why (?) Intel pulled the plug on ATI.
Here are 2 reasons that crossed my mind:
1/ Intel is more stupid than I thought
2/ ATI intentionally instigated an argument with Intel, so Intel would pull the plug, while the back door is open for AMD.
If the second reason has any merit, then the AMD/ATI merger is very stinky, and I have no intention in wearing a gas mask just because of AMD.
Score: 0
nVidia has always been the best for compatibility, and if they keep the same business model, they can still compete for the AMD market, despite the ATI-AMD merger. I doubt nVidia would want to merge with Intel. Intel already has its own onboard graphics division.
Score: 0
I have to agree with Pixelmack... Until I see real world results, its all speculation. It looks promising though.
Score: 0
Blah-blah-blah...you gonna bark little doggy or are you gonna bite?
Score: 0