AMD reveals plans for 25x efficiency gains by 2020

Jun 22, 2014 by Nancy Owano weblog

(Phys.org) —California-based semiconductor product company, AMD, has announced goals for very ambitious efficiency gains in its products over the rest of the decade. AMD has already made headway in improving the energy efficiency of its products by more than tenfold during the last six years (2008 to 2014). Now, in a Thursday announcement, AMD revealed the new intent which involves a "25X20" target, namely AMD intends to realize a 25x improvement for their processors over the next six years. "Through APU architectural enhancements and intelligent power efficient techniques, our customers can expect to see us dramatically improve the energy efficiency of our processors during the next several years. Setting a goal to improve the energy efficiency of our processors 25 times by 2020 is a measure of our commitment and confidence in our approach."

APUs refer to Accelerated Processing Units; AMD combines CPU and GPU compute cores and special purpose accelerators such as and video encoders on the same chip in the form of APUs. This innovation from AMD saves energy, said the company, by eliminating connections between discrete chips, reduces computing cycles by treating the CPU and GPU as peers, and enables the seamless shift of computing workloads to the optimal processing component.

The 2020 goal was announced in Dalian, China, on Thursday, at the China International Software and Information Service Fair, running from June 19 to June 22. Mark Papermaster, senior vice president and chief technology officer, referred to an "attending relentless focus on " as part of AMD's business strategy.

One of the focus areas will be on what AMD called intelligent, real-time . Idle time is the interval between keystrokes, touch inputs or time reviewing displayed content. Executing tasks as quickly as possible to hasten a return to idle and minimizing the power used at idle is important, said AMD, for managing energy consumption. "The latest AMD APUs perform real-time analysis on the workload and applications, dynamically adjusting clock speed to achieve optimal throughput rates." AMD also offers platform-aware power management where the processor can overclock to quickly get the job done, then drop back into low-power idle mode.

Lastly, AMD will leverage its R&D strengths. "AMD recognized the need for energy efficiency years ago and made the research investments that have since led to high impact features." Going forward, said the company, "differentiating capabilities such as inter-frame power gating, per-part adaptive voltage, voltage islands, further integration of system components, and other techniques still in the development stage should yield accelerated gains."

TIRIAS Research analyst Kevin Krewell, commenting on AMD's paths to pull off the target of 25X20, said, "We believe that AMD will achieve its energy efficiency goal, partially through process improvement but mostly by combining the savings from reducing idle power, the performance boost of heterogeneous system architecture, and through more intelligent power management."

Explore further: Car hacking: The security threat facing our vehicles

More information: www.amd.com/en-us/press-releas… nergy-2014jun19.aspx

add to favorites email to friend print save as pdf

Related Stories

AMD's Trinity is out to rattle Intel's Ivy Bridge

May 15, 2012

(Phys.org) -- AMD has announced Trinity, its second-generation A-Series accelerated processing units (APUs), which are out to rival Intel’s Ivy Bridge processors. AMD’s Trinity is an update to its ...

AMD unveils first-ever 5 GHz processor 

Jun 12, 2013

AMD today unveiled its most powerful member of the AMD FX family of CPUs, the world's first commercially available 5 GHz CPU processor, the AMD FX-9590. These 8-core CPUs deliver new levels of gaming and ...

AMD Fusion APUs plumped, primped, stretched

Aug 23, 2011

(PhysOrg.com) -- AMD has announced some new additions to its Fusion APUs optimized for notebooks, netbooks and PCs. Users will see the immediate difference in battery life and enhanced graphics. So what’s ...

AMD whets Steamroller appetite at Hot Chips

Aug 29, 2012

(Phys.org)—Some might think of AMD as the chip design market underdog, but that underdog raised interest this week when the company's CTO, Mark Papermaster, Senior Vice President & Chief Technology Officer, ...

Recommended for you

Yelp to pay US fine for child privacy violation

43 minutes ago

Online ratings operator Yelp agreed to pay $450,000 to settle US charges that it illegally collected data on children, in violation of privacy laws, officials said Wednesday.

Researchers find parking space solution in PocketParker

3 hours ago

Looking for a parking spot? Circling round and round in a lot, feeling the heat of no luck no matter where you look? Could smartphone-tracking movements be of any help? Caleb Garling in MIT Technology Review ...

User comments : 14

Adjust slider to filter visible comments by rank

Display comments: newest first

verkle
3.7 / 5 (3) Jun 22, 2014
That is almost 2x increase in effeciency every year. Better than Moore's law.
TheGhostofOtto1923
3.4 / 5 (5) Jun 22, 2014
That is almost 2x increase in effeciency every year. Better than Moore's law.
Nope.

"Moore's law is the observation that, over the history of computing hardware, the number of transistors in a dense integrated circuit doubles approximately every two years."

-The article is about energy efficiency which has nothing to do with moores law. You must ask yourself what else do you think you know, that in reality you don't.

Or better yet you may want to assume that you routinely don't know what you are talking about, and just look everything up before you post.
IamVal
3 / 5 (2) Jun 22, 2014
I don't know where you got your quote from otto, As you gave no source. The true definition of moores law also includes halving in size.

and since has been extrapolated (not directly from the original law) into gains in processor speed, ram size, power efficiency, bandwidth, read and write speed, and going as small as gate size....
where these measures also seem to double in roughly the same time.

where as moore didn't himself extrapolate to these conclusions, Inferring a relation to the original conjecture due to the similarity in factor and time is not even close to the egregious error you seem to think it is.

Eikka
5 / 5 (2) Jun 22, 2014
The true definition of moores law also includes halving in size.


No it doesn't.

Gordon Moore, Electronics Magazine 19 April 1965.
The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such a large circuit can be built on a single wafer.[1]


The Moore's "law" is meaningless because the doubling interval has been changed at least three times by now, and nobody can agree exactly what it's supposed to measure.

where these measures also seem to double in roughly the same time.


They don't.
TheGhostofOtto1923
5 / 5 (1) Jun 22, 2014
don't know where you got your quote from otto, As you gave no source
If you were used to looking things up you would know how easy it is to drop quotes in google to find the source. But the def is a standard one. Look it up.
The true definition of moores law also includes halving in size
No, it doesn't. Look it up.
Maggnus
5 / 5 (2) Jun 22, 2014
]Nope.
"Moore's law is the observation that, over the history of computing hardware, the number of transistors in a dense integrated circuit doubles approximately every two years."

-The article is about energy efficiency which has nothing to do with moores law. You must ask yourself what else do you think you know, that in reality you don't.

Or better yet you may want to assume that you routinely don't know what you are talking about, and just look everything up before you post.
Otto, you can be an ass, but I admit you make me chuckle at times! Blunt, but true.
IamVal
not rated yet Jun 22, 2014
that quote, while not specifically stating it, implys the idea of halving quite noticeably.
the last line being the key.
If the size weren't to have been improving at the same rate, fitting 65000 transistors on a 'single wafer' then would put your wafer at around half a square mile.

what i'm saying is that the essential framework, factorial (exponential) improvement, over some timeframe, regardless of the subject makes it fair game to directly compare to moores law.
he's not said 'moores law says, so we're doing better than we'd hoped', he said, the improvement rate exceeds more's laws framework of doubling every 2 years.

and excuse my misstep of saying that they're following the same rates, that isn't true. But they are growing exponentially, and a doubling does, by definition, take some interval, so it can be directly compared to the very well known predictions of 'moores law' which is a colloquial name, and not a scientific title, whether positively or negatively.
TheGhostofOtto1923
2.5 / 5 (2) Jun 22, 2014
"Otto, you can be an ass, but I admit you make me chuckle at times! Blunt, but true."

-I don't think casual and lazy posters don't realize what asses they can be. I think we have a responsibility to get things RIGHT and not perpetuate misconceptions.

This is a science forum not some freeking chat room. And it is easy to research and post quotes from experts rather than pretending that you are they (not you personally Magnuts you are always conscientious if not right).

"makes it fair game to directly compare to moores law."

-No it doesn't. It makes it 'fair game' to attribute it to something more appropriate. Look up the source and see if you can figure out what that might be.

"'moores law' which is a colloquial name, and not a scientific title"

-LOOK it up and SEE if you can find an expert who agrees with you, and then POST what he says. If you can't then ADMIT you made it up.
TheKnowItAll
2 / 5 (1) Jun 22, 2014
This is nice but all that tells me is that they've given up on increasing core computing speed. Next will be to work hard on inventing a proper multiprocessor architecture and an operating system to go with it.
Rob
not rated yet Jun 22, 2014
Bickering aside I want to know how much of this they project to come from process node improvements (Moore's Law applies to this) and how much is architectural. I am guessing they get as substantial reduction going from 20nm planar - 14nm finfet - 10nm - 7nm and they don't even work on this any more. That will be GlobalFoundries and the IBM Albany alliance that does that. So what I am asking is how much of this is BS PR and how much is solid engineering?
KBK
5 / 5 (4) Jun 22, 2014
would you dickweeds stop arguing and try to get on with having a good life, instead of wasting it with this abominable shit.

godddamn internet crap throwing monkeys sure know how to ruin the world.

Sure can't fix it... but they sure know how to whine.

If that is the case then keep it to yourselves please, and move on.

If you can't be politely constructive..then...Go someplace else.
alfie_null
not rated yet Jun 23, 2014
Targeting desktop machines? (not a growing segment)

Smart phones?

Smaller embedded devices?

Server farms? Which then makes me wonder how they anticipate these gains will work with VMs?

Entirely in hardware, or do they anticipate tweaking the OS?
TheGhostofOtto1923
1 / 5 (1) Jun 23, 2014
"would you dickweeds stop arguing and try to get on with having a good life, instead of wasting it with this abominable shit"

-You mean abominable shit like this cowpie of yours kbk?

"... all we're doing is forcing the realization of the geometric vibrational intrusion/alignment/polarization into this space... the basics (electrons/positrons) is probably all that could be enabled/gated/resonated/polarized/vectored/spun into occurring, or more properly, gated into being realized into this space/time of quanta we call 'the universe. (In all our/it's limited glory"

-Verkle, kbk returners and many others think physorg is bullshit central where brainless chitchat and creationist proselytism will go unaddressed.

Theyre mistaken.
IamVal
not rated yet Jun 26, 2014
all I was saying is that it is not a law in the scientific sense.. It's a conjecture based upon observation and while being generally accurate for the timeframe proposed by the original postulation, It has continued at a much slower pace since.

no 'expert' in their right mind would say that moore's conjecture is a law such as gravity, entropy or the speed of light.

at this point I'm obviously arguing semantics. So, at that note, Untill it comes up with some empirical type argument where opinion or interpretation means nothing, I'll be making my exit here. but the good and bad thing here is that we generally don't disagree. I see where you're coming from... it's like you're a protestant and I'm an agnostic. Our differences are trivial, why should we make the entire encounter about them?.
just agree to disagree. in this case, it's not like it'll have any ramifications what so ever.

I'll give you 'halving', should have put something else as halving is too specific.