News
 

Bookmark and Share

(2) 

Standalone central processing units (CPUs) for client computers will be challenged by accelerated processing units (APUs) with integrated graphics core, a high-ranking officer at AMD said in an interview. APUs provide better performance because of its ability to efficiently process both parallelized and serialized data.

"I think APUs will definitely challenge standalone CPUs. I believe that the future of consumer as well as commercial computing environments are characterized by the ability to present a compelling visual experience. Taking a GPU core and a CPU core and using them together on one chip will definitely challenge standalone CPUs," said Neal Robison, senior director of content and application support, in an interview with X-bit labs published on Sunday.

Accelerated processing units integrate many parallel GPU processing elements as well as several "fat" x86 CPU cores that can efficiently process typical data. Many performance-demanding applications nowadays use parallel processing and receive benefits from both multi-core CPUs as well as many-core CPUs. Nonetheless, there are loads of programs that use advanced microprocessors and have no need for GPUs.

But while APUs will confront low-cost systems, they will not challenge and will not become part of advanced personal computers with discrete graphics cards.

"I do not think that APUs will challenge discrete GPUs on anything, but on the lowest-end systems. When you look at adding a discrete GPU that enhances performance of the graphics side, it makes a huge amount of sense as it scales [performance] on a wide amount of applications because of the rich visual experience that everybody expects now when they are actually using their computing device," added Mr. Robison.

 

Tags: AMD, ATI, Fusion, Phenom, Radeon, Llano, zacate, Ontario, Wichita, Zambezi

Discussion

Comments currently: 2
Discussion started: 03/20/11 02:43:40 PM
Latest comment: 03/22/11 07:33:28 PM
Expand all threads | Collapse all threads

[1-1]

1. 
If you look at the Radeon cores that are being used several things really stand out. The HD 6370 is a 7 watt core as discrete GPU so they probably draw much less as APU. Clearly this is headed for laptops. These cores also have one other HUGE common denominator.

Here's a link for Radeon releases: http://en.wikipedia.org/wiki/Compa [...] sing_units

They were all released Nov 2010!!! If anybody does not believe that AMD Llano will eliminate the mid price point mass market for discrete gpu's then they really need to have a hard look at the facts.

AMD is releasing Radeon 6990 without quantity restrictions. Nvidia is releasing less than 1000 GeForce 590's. The 590 is a cherry picked dual gpu board. It may perform equal to or actually outperform Radeon 6990. But what good is it if you can't buy it? Or is the market for bleeding edge bragging rights also just not there?

The mass market supports new gpu core development. Without the sales of millions of discrete gpu's for legacy upgrades, the next generation doesn't get designed or if it does without the prospect of any mass sales volume then it becomes a very expensive piece of silicon.

A good example is the ATI FirePro and Nvidia Quadro brands. They simply do not have the mass volume sales to allow for a lower purchase price point of $2000-$3000.00, the demand is simply not there. Product refreshes are also not as frequent as the mass market again due to demand.

If AMD is using this year’s top discrete gpu design for next years Fusion APU then the discrete gpu market is most certainly dead. Will there be a reason to upgrade a one year old Llano box with the latest discrete GPU? For what gain other than bragging rights? And what would be the discrete GPU demand looking forward?

The real question becomes is that AMD’s plan? And if so how does Nvidia plan to keep the discrete market open? Does Nvidia license core designs to Intel?

The other question is just what does AMD plan to do with Bulldozer? It seems that Bulldozer will be the server, workstation or high performance desktop and gamers cpu. This is certainly not a mass market cpu. As a server obviously graphics are not needed beyond a motherboard integrated gpu. So there will be some demand for discrete gpu boards with Bulldozer.
The next question becomes. When does AMD release Bulldozer with an on die graphics core? Because Bulldozer will be the only market left open for discrete gpu’s.

Of course just how Intel intends to answer AMD will determine the future of Nvidia graphics. Arguably Intel cannot compete with the AMD/ATI library. Every few months AMD releases new graphics silicon, they are continually evolving that product to meet present market demand. Intel is not a graphic’s design house. But now they have to be to keep their CPU business competitive. That means they are designing graphics gpu’s to penetrate a market that is owned 100% by AMD and Nvidia.

AMD is now designing discrete GPU’s with the intention of integrating that design on-die for an APU release ONE YEAR LATER! That has to be an optimized model and as such just how can Nvidia compete with AMD if they don’t have that insight into Intel future release Architectures? Nvidia’s only market will be on an Intel Inside box.

Right now AMD is directing the future of CPU design. They have the edge over Intel with ownership of arguably the world’s best graphics design portfolio and gpu design team. And they have the cost edge over Nvidia as they simply sell a one year old core design on die to millions of consumers as an APU. For Intel to remain competitive they are forced into the same model and this model shuts out Invidia.

..
0 0 [Posted by: akamateau  | Date: 03/20/11 02:43:40 PM]
Reply
- collapse thread

 
Your logic is flawed akamateau(Cough: "AMD fan boy" lol). Your wiki link goes no where. Right now AMD is not directing the future of CPU X86-X64 design, Intel is for the X86-x64 architectures. Intel releases cpu architectures every 2 years, then does a die shrink every year. How often does amd releases brand new architectures , and does die shrinks? Not that often.

For the amd releasing architecture part, until their processors are no longer competitive with Intel's processors. Look at current AMD processors, & Nehalem, Ya i'm talking about the later generation of Intel. An intel quad core can kick an amd 6 core butts hands downs in most, if not all benchmark & it gets better with sandy bridge. Just look at these articles "CPU Benchmark, Part1, 2 , & 3", from Xbitlabs. Intel got recommend buy in part 3 & editors choice in part 2 for being the fastest. & AMD just got recommended buy in part 1 & 2. In articles 3 & 2 Intel kick but. We all now why, that is. Intel rocks for raw performance. Amd is cheap because Intel beats them in most performance matches. Ya i know about price, for the most part amd is cheaper because "They can not keep the pace with that of Intel for the moment & it looks like its only getting worst for AMD.

For the die shrink part. AMD was flat out "stupid" for selling its fabs & out sourcing to different companies. Thats why they are no longer competitive with Intel for die shrinks. AMD Lacks Most if not all control over the fab shrink process.

If you look at all the supported instruction sets. Intel has more supported instructions sets. Just look at these 2 wiki Lists & compare Sandy Bridge & AMD fusion for instruction sets

Sandy Bridge List "http://en.wikipedia.org/w...idge.22_.2832_nm.29"

AMD Fusion List "http://en.wikipedia.org/wiki/AMD_Fusion"

You will notice that all of MMX & 3D-Now instruction sets are gone except for 2 3D Now instructions here's the article for proof "http://www.xbitlabs.com/n...w_Instructions.html"


If you look at the AVX instruction set for both Intel & AMD. AMD screwed themselves from the get go, with their "Flex FPU". If those 2 integer core issues a total of 2 or an even number of AVX instructions, Intel's processors will crush AMD. In AMD case "Its like hyper-threading for the FPU". Its gets even worst for AMD with Intel Hyper-threading technology enabled, because with 2 128-bit FPU instructions that one 256-bit fpu acts just similar to, or just like AMD "Flex FPU".

Here's another point, count the number of supported features leaving out ram speeds, comparing fusion with sandy bridge, from the 2 wiki articles

You will find out the Fusion supports 9 features, 10 if you are counting GPGPU computing, & amd could use that GPGPU computing for video trans-coding too, just like Nvidia.

Now look at Sandy Bridge its supports 18 Features, 19 if using either AMD/ATI or Nvidia GPGPU features. Double the amount that AMD supports if not more. just using the Wiki Articles.

Now for the GPU/APU front.

Intel loses for the GPGPU front i think ??? For video trans coding Intel doesn't lose. Because according to this xbitlab article about the sandy bridge architecture. "http://www.xbitlabs.com/a...ecture_5.html#sect0" ;
Sandy bridge has a unit for decoding & encoding. AMD only has CPU/GPGPU for encoding, using additional resources from the GPU & or cpu, & a separate unit for decoding. Which equals less efficiency.

For 3D games the intel hd sucks for all category's except for one & that's low end/ entry level/integrated chips. It gets better though if you forgo Intel HD chip. Go with your cpu competitor GPU solution "AMD", or use Nvidia. Either one is a good choice.

My estimates is. If you use either Nvidia or AMD GPU with Intels Builds it just proves that AMD CPU part of its APU's will suck. Whether its thee Brand New "Fusion" or not.

Overall Intel is good for Performance, but sucks for price. AMD is the opposite for the most part, its sucks for performance, but rocks for the price.

After reading this article & the many articles about AMD CPU's, AMD/ATI GPU's, AMD APU's,Intel CPU's, Nvidia GPU's & replying to this post, while going through a bunch of web pages & a maybe a few websites to back some of my points.

AMD is flat out, a spineless, weakling, stupid Cheap Skate, company. which i would enjoy watching the company die a slow painful death if & when they fail.

Reasons why i think this way about AMD
1. Selling their fabs & out sourcing.
2. Not promoting their 3D-now Set, & having to ditch the 3D-Now set with the introduction of fusion APU.
3. Reducing their feature set when compared to Intel, & not having the gumption to go head to head with intel, adding new instruction sets & or features.
4. AMD thinking that their "Flex FP" will do the job. In reality it won't because its 2 cpu cores with an HT FPU for older instructions or in the case of AVX or 256-bit instructions, 1 FPU shared between 2 cpu cores, which means a reduece performance overall. Look at the instruction sets made be intel. AVX will be a standard in the future.
5. Using last years "outdated" GPU's with then current AMD CPUs, just to be price competive competitive with intel.
6. Not being performance competitive with Intel for a long time.
7. "My estimates" Charging a premium for a feature reduce CPU with an "outdated GPU".
8. Have marketing crap for the undereducated saying the amd APU is better taking advantage of people.
9. "AMD Fan Boys" & Stupid people alike thinking that Fusion APUs will compete with Intel/Sandybridge CPU's. Do your research & read


"For People in general"
Read up on all past, current, upcoming & future technology, by any means possible, & read the facts from multiple credible sources

"XBITLABS SUCK WHEN IT COMES TO COMMENTS"
0 0 [Posted by: macintoshci  | Date: 03/22/11 07:33:28 PM]
Reply

[1-1]

Add your Comment

[Login] [Forgot password?] [Registration]




Related news

Latest News

Thursday, May 24, 2012

10:00 pm | Microsoft Clarifies Its Exec's Claims: 500 Million Windows 8 Copies in 2013 Are "Potential" Upgrades. Microsoft Retracts Statement Regarding 500 Million Windows 8 Licenses to Be Sold in 2013

8:35 pm | ECS Reveals "NonStop" Mainboard Family with Extended Stability and Reliability. ECS Boosts Stability and Reliability with Premium Components and Rigorous Testing on NonStop Platforms

6:28 pm | AMD Rumoured to Start Production of Next-Gen FX-Chips in Q3. AMD to Start Making FX "Vishera" Chips Next Quarter

11:32 am | UMC Begins to Build Fab to Make 28nm, 20nm and 14nm Chips. UMC Spends $8 Billion on Expansion, Confirms Development of 14nm FinFET Process Technology

Wednesday, May 23, 2012

11:17 pm | OCZ Quietly Introduces 7mm Vertex 3 LP SSDs for Ultrathin Laptops. OCZ Reveals Low-Profile Vertex 3 LP Solid-State Drives

11:01 pm | Toshiba Vows to Release Hybrid Hard Disk Drive in September. Toshiba Confirms Development of Hybrid Hard Disk Drive

10:28 pm | Jury Finds: Google Android Did Not Infringe Oracle's Patents. Google Not Guilty of Infringing Oracle's Patents

8:48 pm | 208 Million "Phablets" Set to Be Sold by 2015 - Analysts. Smartphones with Ultra Large Screens Will Continue Gaining Popularity

8:23 pm | Microsoft Expects 500 Million People to Use Windows 8 Next Year [UPDATED]. Microsoft Projects Breakthrough for Windows 8 Sales Thanks to Compatibility with Different Devices

7:07 pm | Thunderbolt Interface to Receive Green Light from Mainboard Makers in Second Half of 2012 - Rumour. Thunderbolt Interconnection to Make It to Mainstream Systems Later in 2012

6:18 pm | Apple and Samsung Fail to Reach Settlement at Special Meeting. Court-Ordered Meeting Does Not Resolve Apple Vs. Samsung Dispute

12:10 pm | Seagate Set to Take Over LaCie. Seagate and LaCie Agree About Acquisition