AMD Piledriver Cores to Use Energy Recovery Tech

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

SteelCity1981

Distinguished
Sep 16, 2010
1,129
0
19,310
The problem is people tend to focus on what a CPU can do instead of what a CPU can do for you. Benchmarks mean little to most consumers because they will never play the most demanding games or render the most demanding graphics etc... Most consumers couldn't tell the diff between an Athlon X2 and a Core i7, because most consumers don't need that much power to tell the diff, but what most consumers do need is a CPU that is conservative on power. As more and more mobile device are used more and more consumers will look at the most important benchmark to them and that's power consumption which will be the key to technology going forward and not so much how fas an CPU can render and object or how many fps an CPU can handle. This is the Reason why Llano is a success and this will be the reason why Trinity will be even a bigger sucess. Times are changing We've gone from CPU power to now how much can a CPU conserve power.
 

omega21xx

Distinguished
Jan 21, 2012
863
0
19,060
[citation][nom]mr_wobbles[/nom]I still dont see why people whine and complain about how Bulldozer was awful. If you want to start Intel fanfag-raging go ahead. If you guys havent noticed, not all PC gamers have 1500$+ rigs. I WISH I could have something as epic as Bulldozer in my laptop, but my current CPU (AMD Phenom X4 N950 @ 2.1 ghz) doesnt get maxed out in any game I have, and Ive got BF3, Anno 2070 and many others. I see no reason why you would spend the 80-110$ more on an i5 2500K than on a Zambezi , which my friend has and hasnt had a single problem. GO AMD!!! Lets hope for a good year from AMD.[/citation]
The biggest problem people fail to realize, is not that it's bad because it doesn't compete with sandy bridge cpu's, it's that it is out performed more often than not be it's former generation, phenom ii.
 

alidan

Splendid
Aug 5, 2009
5,303
0
25,780
[citation][nom]omega21xx[/nom]The biggest problem people fail to realize, is not that it's bad because it doesn't compete with sandy bridge cpu's, it's that it is out performed more often than not be it's former generation, phenom ii.[/citation]
it only fell behind the phenom II because of single core applications... which is honestly BS because for at least 5 years we have had dual core as a standard, correct? why are things still being made for single core only?
 

r0ck3tm@n

Distinguished
Sep 27, 2009
136
0
18,690
[citation][nom]omega21xx[/nom]The biggest problem people fail to realize, is not that it's bad because it doesn't compete with sandy bridge cpu's, it's that it is out performed more often than not be it's former generation, phenom ii.[/citation]

Most people who use a computer will believe what they are told by knowledgeable people. From AMD themselves down to knowledgeable enthusiasts like us down to those who just use a computer for work, many of us know that the Bulldozer architecture was a disappointment. I wish I knew all the reasons for this, mostly I wish I knew whether the architecture itself is faulty or more so the manufacturing process.

The important thing to come to realize is that the first production of that architecture is behind us. It seems to me that perhaps AMD is continuing to work on it. I hope so. Because what I do not want is to come visit Tom's Hardware and see no mention of AMD because there is no AMD.
 

DRosencraft

Distinguished
Aug 26, 2011
743
0
19,010


It says at the end of the article that AMD is aiming for 5-10% efficiency uptick with piledriver. I don't think that is a particularly bold claim, imbedded tech aside. I can't say much about how the tech works, but I can't see 5-10% as a particulary exagerrated claim. Yes we will need to see how it actually pans out. Assuming the technology works, the target 5-10% TDP drop (which would still have the 8-core Zambezis at over 100W) could be a realtively conservative estimate.
 

juan83

Distinguished
May 17, 2011
53
0
18,640
[citation][nom]alidan[/nom]it only fell behind the phenom II because of single core applications... which is honestly BS because for at least 5 years we have had dual core as a standard, correct? why are things still being made for single core only?[/citation]

because sometimes is faster, more eficient, and take way less time to code a single threaded applications than multithreaded. It's a pain in the ass write code wich taxes all cores..

for example try to compress a huge file with 7-zip and set the dictionary size = 1024 mb. it takes more than 20Gb of ram with 4 thread (4 cores). the speed is sooooooooo low due swapping that you ll gonna cancel and restart it with 2 or 1 thread, then the hole work is done in ram and the speed is faster. more over, your pc will respond meanwhile.

Luckily openCL is emergin, where you can tax thousands of cores from graphic cards (and higher bandwidth)
 
Such a lot of marketing.... such a lot of debate, luckily we live in a non-democratic world but in a free market. So we can choose what we want to buy without being brain washed or intimidated by the fan-fags all around.
 

r0ck3tm@n

Distinguished
Sep 27, 2009
136
0
18,690
[citation][nom]alyoshka[/nom]Such a lot of marketing.... such a lot of debate, luckily we live in a non-democratic world but in a free market. So we can choose what we want to buy without being brain washed or intimidated by the fan-fags all around.[/citation]

Fan-fags? Fanboy doesn't work for you?

Basically, you are whining that people are here talking about the article? I think you need a beer.
 
10% power consumption reduction, 10 % increase in clocks and a 10% increase in efficiency, that itself is nearly a 30% change from BD to PD :) Way to go AMD. You had that timeline for 4 years all the way to the Excavator, glad to see if it happens within a year....
 

shinku_sai

Distinguished
Jun 8, 2009
35
0
18,530
Maybe... AMD should ALSO INCORPORATE the V6 engine on the Nissan GTR R-35 instead of this non sense..... Yeah... They got that "TECH" from a car... Well seems like you are look at the WRONG part of the car that you should be getting "TECH" from....
 

omega21xx

Distinguished
Jan 21, 2012
863
0
19,060
[citation][nom]alidan[/nom]it only fell behind the phenom II because of single core applications... which is honestly BS because for at least 5 years we have had dual core as a standard, correct? why are things still being made for single core only?[/citation]
The problem being that per core performance largerly hasn't changed in years until sandy bridge came about, adding more cores wouldn't decrease per core performance, the lower per core performance is due to the new arc. and all the problems they brought with it. When I made a statement earlier, it was in terms of saying a phenom ii x4 at the same clock as a FX-4100 or phenom ii x6 vs a FX-6100. Now when you add in more cores as in the 8120 then you really shouldn't compare the two on multithread performance. Either way my point being that the per core performance should not have dropped, AMD had made a statement saying something along the line of due to the arc. they would have to increase clocks 30% to reach a higher performance than the phenom ii's. Obviously they couldn't overcome this as none of them come with 30% increase in clocks over what phenom ii currently offers.
 

apus are okay for now. they don't have bulldozer-based cpu cores inside yet. amd did achieve power efficiency with their 6000 and 7000 series cards, and mobile apus. they should have achieved it with bd too, but they didn't - that's what i'm concerned about.

you can configure an intel pc within $600-$800 and shame a similarly priced amd gaming pc in terms of both gaming performance and power efficiency.
i hope you get to buy a zambezi (preferably one with fx 8150) laptop. :D

if amd can lower even 2-3% load power consumption, it'd be huge considering how power hungry zambezi is. afaik tdp isn't the same as power consumption or efficiency. tdp is just a measure of nominal power consumption at a typical (lab controlled) workload and it is very brand-specific i.e. different companies put different workloads on their cpus before assigning tdp to them.
 

triny

Distinguished
Feb 2, 2012
450
0
18,790
It's like +.4 ghz for free I imagine it would revolutionize ulv designs as well give that much more head room for OCing
 

wiyosaya

Distinguished
Apr 12, 2006
915
1
18,990
I am sure I'll get voted down for this one, however, I cannot help but wonder just how much of this is marketing blather.

Sorry AMD. I'll keep an eye on your developments, however, all this marketing oriented "talk" is wearing thin on me these days. Stop blathering about the great things you are doing AMD, and actually release a great CPU.
 

wiyosaya

Distinguished
Apr 12, 2006
915
1
18,990
[citation][nom]Shin-san[/nom]Not bad. Hope this means that AMD will be competitive again[/citation]
IMHO, it is going to take much more than this. This, as I see it, is a trivial improvement.
 

f-14

Distinguished
[citation][nom]chumly[/nom]But will it best a 2500k?[/citation]
no it will not, the aim is to reduce power usage while achieving high clock rate, not in actually increasing the instructions per clock.
also pile driver sounds to be a server only chip.
 

triny

Distinguished
Feb 2, 2012
450
0
18,790
big gains in power per watt ,increase in clock speed
the prevailing winds buffeting AMD are strong, AMD literature claims Piledriver cycle is all about
lower power consumption and higher clocks,the cyclos resonant web should help in that regard immensely.
I am pulling for them we need more competition for Intel it's the mother of invention.
 
G

Guest

Guest
Was looking for a general comparison of Piledriver vs Bulldozer and found this today: http://blog.netflowdevelopments.com/2012/05/02/amd-bulldozer-vs-amd-piledriver-one-amd-apu/

Covers a lot of what's been talked about here with few more tidbits
 
Status
Not open for further replies.