Monday, August 31st 2009

AMD Announces Six-Core Opteron EE Processor with 40W ACP

AMD today announced the immediate availability of the new Six-Core AMD Opteron EE processor at 40W ACP. Delivering up to 31 percent higher performance-per-watt over standard Quad-Core AMD Opteron processors, the Six-Core AMD Opteron EE processor is tailored to meet the demands of customers who need strong performance, but must trim out every watt possible in a server system and reduce the datacenter’s power draw.

  • Many customers requiring this very low power processor deploy dense, large-scale IT projects where system power trumps raw performance and every watt of power savings can have significant positive impact on the bottom line
  • AMD offers a wide range of energy-efficient processors; customers now have 15 low and very low power server processor choices available
  • All of AMD’s low power processors are fully-featured and available with the same AMD Virtualization (AMD-V) technology and AMD-P features for advanced virtualization and power savings capability as standard power versions
  • This 40W ACP processor is currently being integrated in custom solutions programs and will be available from system builders for cloud and Web 2.0 customers where density and low power are key considerations
“It’s important for our industry to look at how technology is being used and where customers have emerging needs,” said Patrick Patla, vice president and general manager, Server and Workstation Division. “The AMD Opteron EE processor enables OEMs to deliver energy-sipping servers that don’t compromise on power management, virtualization or performance features. It is specifically designed to help address the challenges that are generating a great deal of discussion these days - building and running very dense data centers for Web services, while doing more with less.”
Add your own comment

25 Comments on AMD Announces Six-Core Opteron EE Processor with 40W ACP

#1
1Kurgan1
The Knife in your Back
Very nice, wish I had the money to build a server :(
Posted on Reply
#2
Mussels
Moderprator
6 core, 40W? hell yeah :D
Posted on Reply
#3
btarunr
Editor & Senior Moderator
They can use the same silicon to make a Phenom II X6 3.20 GHz processor with 140W TDP, if they want to.
Posted on Reply
#4
FreedomEclipse
~Technological Technocrat~
by: 1Kurgan1
Very nice, wish I had the money to build a server :(
but what would you build a server for?? not all servers take expensive hardware to run
Posted on Reply
#5
Sihastru
by: Mussels
6 core, 40W? hell yeah :D
From AMD's website:
Q: How does ACP relate to Thermal Design Power (TDP)?
A: In the simplest terms, ACP conveys what a user’s CPU power experience may be like while running workloads under typical data center environmental and operating conditions. In contrast, AMD’s TDP is the processor design specification. The two can be quite different as a result. AMD’s TDP and ACP values are intended for completely different uses:
* TDP is for system designers.
* ACP is for people who want to know what the CPU power may be like under real-world conditions.
The ACP value can have relevance for people in job roles that require them to estimate power consumption for the purpose of power budgeting. AMD will continue to provide thermal design power (TDP) specifications to platform designers and industry partners in the AMD thermal and power datasheets.

Q: Is ACP replacing TDP?
A: No. ACP and TDP will coexist.
So, like the "PR" rating of Athlons and Durons a few years ago, ACP is designed to make a CPU seem better then it actually is. So a 40W ACP could translate into an 125W TDP... maybe not that much... anyway ACP < TDP. Looks like marketing because lately the power efficiency game they played 6 months ago came back to bite them in their behinds.

Now Average Joe (and some of you already) will see then wattage and not see the ACP that has no meaning for him, and think that the AMD 40W ACP processor is much more power efficient then the 65W TDP Intel processor... See?
Posted on Reply
#6
Mussels
Moderprator
bugger :(
Posted on Reply
#7
Kitkat
so still super hell yeah! lol!
Posted on Reply
#8
Sihastru
The TDP for this processor is rated at 60W so for an AMD processor this is pretty good. Clockspeed is 1.8GHz. Again, AMD's TDP and Intel's TDP are not really directly comparable, but compared to an 140W desktop quad core this seems great.
Posted on Reply
#9
Mussels
Moderprator
by: Sihastru
The TDP for this processor is rated at 60W so for an AMD processor this is pretty good. Clockspeed is 1.8GHz. Again, AMD's TDP and Intel's TDP are not really directly comparable, but compared to an 140W desktop quad core this seems great.
the low TDP makes sense with the low clockspeed - 60W for 1.8GHz, 120W for 3GHz ish?


its not bad really, its a great choice in the server environment.
Posted on Reply
#10
btarunr
Editor & Senior Moderator
by: Sihastru
Now Average Joe (and some of you already) will see then wattage and not see the ACP that has no meaning for him, and think that the AMD 40W ACP processor is much more power efficient then the 65W TDP Intel processor... See?
Uhh ~80W ACP is ~125W TDP. You do the math.
Posted on Reply
#11
MohawkAngel
by: Sihastru
From AMD's website:


So, like the "PR" rating of Athlons and Durons a few years ago, ACP is designed to make a CPU seem better then it actually is. So a 40W ACP could translate into an 125W TDP... maybe not that much... anyway ACP < TDP. Looks like marketing because lately the power efficiency game they played 6 months ago came back to bite them in their behinds.

Now Average Joe (and some of you already) will see then wattage and not see the ACP that has no meaning for him, and think that the AMD 40W ACP processor is much more power efficient then the 65W TDP Intel processor... See?
It was the same during the government laws for lowering horsepower numbers on cars at the end of 70'S. During the 50' and 60's the horsepower advertised by GM for exemple was the raw horsepower of the engine was 450hp. But in fact the net horsepower at the wheels was 220hp. There was raw theorical horsepower on paper, the BHP (brake horsepower) it was the horsepower at the flywheel net after the calculated it on a bench. It lost horsepower because of the internal friction, heat, metal dilatation etc. Finally the net horsepower at the wheel compared to the BHP because of the same heat, friction, dilatation in the transmission, the driveshaft etc.

All this text just to say..Advertisement is for catching up people who don't know. Sales are easier when you foul ignorant people than when you informs people who know the subject because theres more ignorants than people who knows. Also too much technics in an advert it look like a brag to the "mass" people. :ohwell:
Posted on Reply
#12
Flyordie
by: btarunr
They can use the same silicon to make a Phenom II X6 3.20 GHz processor with 140W TDP, if they want to.
If my 2 Istanbuls can do 3.0Ghz @ 100W each... they can do better than 140W @ 3.2Ghz.

Besides that, you are dead on.
Posted on Reply
#13
Mandown
I think I saw this on TPU or maybe somewhere else, That this would be a 40w ACP and 75w or 95w TDP. I have trouble remembering 7's from 9's.:ohwell:
Posted on Reply
#14
A Cheese Danish
If I could snag a single proc for a build, that would be epic!
Posted on Reply
#15
lemonadesoda
by: btarunr
They can use the same silicon to make a Phenom II X6 3.20 GHz processor with 140W TDP, if they want to.
Oh. Does that mean this 40W processor operates at about 1.20 GHz then?
Posted on Reply
#16
btarunr
Editor & Senior Moderator
by: lemonadesoda
Oh. Does that mean this 40W processor operates at about 1.20 GHz then?
Opteron 2419 EE runs at 1.80 GHz. Apparently 40W ACP = 60W TDP. ACP and TDP don't have a direct math relation.
Posted on Reply
#17
thezorro
by: btarunr
Opteron 2419 EE runs at 1.80 GHz. Apparently 40W ACP = 60W TDP. ACP and TDP don't have a direct math relation.
that's incorrect, amd says that acp and tdp are comparable.
the best way to know, is to measure the consumption and see how efficient it is
Posted on Reply
#18
Wile E
Power User
by: MohawkAngel
It was the same during the government laws for lowering horsepower numbers on cars at the end of 70'S. During the 50' and 60's the horsepower advertised by GM for exemple was the raw horsepower of the engine was 450hp. But in fact the net horsepower at the wheels was 220hp. There was raw theorical horsepower on paper, the BHP (brake horsepower) it was the horsepower at the flywheel net after the calculated it on a bench. It lost horsepower because of the internal friction, heat, metal dilatation etc. Finally the net horsepower at the wheel compared to the BHP because of the same heat, friction, dilatation in the transmission, the driveshaft etc.

All this text just to say..Advertisement is for catching up people who don't know. Sales are easier when you foul ignorant people than when you informs people who know the subject because theres more ignorants than people who knows. Also too much technics in an advert it look like a brag to the "mass" people. :ohwell:
You're a little off. They still measured horsepower on a dyno back in the 50's and 60's. It is called bhp because the dyno uses a brake to hold the engine at the desired rpms. At any rate, the reason the numbers are different is because back then, they used to run the engines without any accessories hooked up, like the alternator, power steering pump or even the water pump, etc., etc. When the power values went down, it was because the federal govt mandated that the accessories be installed for the testing, thus lowering the rating of an otherwise identical engine. In 92 or 93 iirc, the govt became even more strict on the testing, mandating that the testing conditions be even closer to real world conditions. Using a Mustang as an example of that change, they went from a rating of 225HP in 92, to 205HP in 93, despite the engines actually putting out the same amount of power. The rating system just changed. To this day, factory horsepower numbers are still measured at the flywheel, not the wheels.

Just a little car history and fyi for everyone. lol.

As far as this cpu, 60W TDP out of a hexacore is still mighty impressive.
Posted on Reply
#19
aj28
ACP really is the proper stat to be giving on these server CPU's. You don't install a server with four CPU's, much less four, six-core CPU's, under the assumption that the monstrosity you've created is going to be running anywhere full-power, all the time, if ever.
Posted on Reply
#21
Steevo
Exactly, and right on the HP rating too. We use a water brake dyno at work to rate engines, and it works well.



And a 60W TDP for full power operation still makes this a awsome chip at the speed, considering most server systems are thread bound, where more cores means more work being done, or the ability to assign cores to workloads means more than processing speed.
Posted on Reply
#22
rickss69
And there is not a dyno on the planet that can measure a top fuel engine...:D
Posted on Reply
#23
btarunr
Editor & Senior Moderator
by: thezorro
that's incorrect, amd says that acp and tdp are comparable.
the best way to know, is to measure the consumption and see how efficient it is
Very well, explain these numbers in an equation:



Numbers taken from specs sheets of various Opteron models. ACP doesn't share a math relation with TDP (since they're both rated values, not measurements). They're just products of blindfolded primates playing darts at AMD/Intel who failed to properly explain ACP/TDP to analysts.
Posted on Reply
#24
lemonadesoda
"Worlds only 40W ACP for 2P servers"

Well that is BS! Intel L5410/L5420 = TDP 50W, which is 20% less than AMD's 60W TDP.

WHAT A LOAD OF OLD SPIN.

OK, so they are trying to split hairs. They mean only "6 core CPU at 40W ACP". Fair enough. But I bet intel's L5530 which is 4 core + 4 HTT = 8 virtual cores is faster AND is in the same TDP/ACP envelope. So it's all down to multitask performance per watt. Let's see some real numbers...
Posted on Reply
#25
Sihastru
Everyone is getting ahead of themselves. People think Intel and AMD measure TDP differently. The truth is they measure it using the same suite of benchmarks, but, because of differences in architectural design of the CPU's, Intel comes up a little better, because of larger and faster caches. AMD feels this is wrong doing towards them and, even if the big industry players like Intel, Sun, HP and IBM have settled on the suite of SPECpower benchmarks run by a committee of industry players hailing from all those firms (and even AMD), are always coming up with excuses...

Now, that Nehalem and the current Opteron line have the same architectural design (at least if you look at a feature set checklist it seems the same, integrated memory controller, comparable L3 caches etc.) and the results are still better for Intel, AMD really has nothing else to complain about. As a result they "invented" this ACP measuring scheme that unfortunately no one else recognizes.

For a good read I must refer you to INQ's "AMD talks ACP vs TDP again" article. If I may add, it's a rare thing on INQ to talk about anything that might put AMD (or ATI for that matter) on a less then perfect light.
Posted on Reply
Add your own comment