AMD Phenom appears in pricelist

[nemrod]

10W+20W+16W+5W= Still gives 100W for just the CPU rather than 40W or 60W some would feign us to believe. This is ludicrous if you understand electronics and how Penryn was made as you are saying these ≥3GHz parts are 60W/80W but cannot explain why Intel never thought that and decided to label them 130W/136/150 for desktop consumers and marketing instead. You are trying to tell me you know better than 1000+ professional processor engineers and over a few thousand Intel employees yet without evidence to disprove them.

I believe we can stop this discussion. You're right on the fact the power is take after the psu. But even with that you have a lot of data above from different sources showing power of cpu is low. You still thinking full system power is cpu power, we have discuss yours data removing different part to have the power of the cm+cpu. I could have add again that it's a X38 motherboard with an hungry chipset etc... But ok penryn are the worst processor ever made, fugger never show 5.8 GHz on cascade on this kind of baby and so one...

[Clairvoyant129]

Stop dreaming please! From expreview, we can expect that phenom at 3GHz is about equal to QX6850. From fanboys it 's probably far better...
Why do you believe that it would be far below the price of QX6850. Does AMD have again several hundred millions dollars to lose each quater in the future? Have you never understand that AMD offer is always relative to intel offer?

Fanbois don't use reason, don't even bother trying to correct him.

[keiths]

nemrod: "efficiency of psu is at best 77%."

You need to brush up on PSUs. 85+ efficiency PSUs have been around for a long while now.

[JumpingJack]

In case you guys haven't realised, Intel's 45nm TDP is *very* conservative. I mean they are literally throwing out TDPs that are twice what the chip consumes.

Under full load, a QX9650 (130W TDP) consumes under 65W... read some reviews!

http://www.lostcircuits.com/cpu/intel_yorkfield/4.shtml

Measured at the socket, 64.8 Watts.

He goes on to note:

At an estimated 70% efficiency of the motherboard VRM, this puts the raw CPU power consumption at around 50 W.

Jack

[JumpingJack]

yes i love it how you fudge figures.
HD9500WCGDBOX CPU Desktop Phenom X4 9500 (4MB,95W,AM2) box $247.00
HD9600WCGDBOX CPU Desktop Phenom X4 9600 (4MB,95W,AM2) box $278.00
HD9700XAGDBOX CPU Desktop Phenom X4 9700 (4MB,125W,AM2) box $288.00

i estimated thats the 3.0 GHz phenom was around $400 and the QX9650 is $1149 (source)
thats a $749 difference with a 1000 points addition. my opinion intel gave some of the greatest overclockers in the world there chip to take the world record so they could price things higher. for the latest chips the come out AMD is going to take the crown for price to performance.

now if you dont mind intel fan boys look but dont post your smart *** comments.

He wasn't fudging.

http://www.google.com/products?q=Phenom+9700

300-330 bucks... looks like computsource is the place to go.

Consider that from the data we have seen so far, leaked as well as simulated (Anand's preview), a Phenom 9700 < Q6600 (performance), yet for the price the Phenom 9700 > Q6600.

This is not including the 2.5 GHz Yorkfield that if you stretch your imagination (a big stretch considering the source) will come in around 270. The only reason to buy a Phenom is if you have an AM2 board you want to upgrade.

[nemrod]

nemrod: "efficiency of psu is at best 77%."

You need to brush up on PSUs. 85+ efficiency PSUs have been around for a long while now.

No, the legend of the figure says: "PCPC 1KW PSU that gives 77% efficiency at the qualified load." I was wrong removing efficiency because that was already suppress for the figure but the 77% efficiency is stated by the authors of the figure.

[ryboto]

http://www.lostcircuits.com/cpu/intel_yorkfield/4.shtml

Measured at the socket, 64.8 Watts.

He goes on to note:

Jack

how can they measure at the socket?

[mstp2009]

how can they measure at the socket?

They tell you in the article:

"measured before the VRM"


Meaning they actually strapped a potentiometer onto the soldered leads at the motherboard as it was running (or more likely, before they hit the power button).

This is by far the most accurate method for calculating power consumption.

[informal]

BLT in USA will have 3200 of 9600s coming in 4 days in California region.Price 291$.

[mstp2009]

BLT in USA will have 3200 of 9600s coming in 4 days in California region.Price 291$.

That's good news. So which AM2 owner here will get one first and put it through the paces?

[nemrod]

how can they measure at the socket?

For this one:
http://www.matbe.com/images/biblio/a...0000063320.png

They measure current deliver to the cpu with something called a "pince ampérométrique." I believe translation is "clamp-on ammeter"

http://en.wikipedia.org/wiki/Clamp_meter

[knightwolf654]

He wasn't fudging.

http://www.google.com/products?q=Phenom+9700

300-330 bucks... looks like computsource is the place to go.

Consider that from the data we have seen so far, leaked as well as simulated (Anand's preview), a Phenom 9700 < Q6600 (performance), yet for the price the Phenom 9700 > Q6600.

This is not including the 2.5 GHz Yorkfield that if you stretch your imagination (a big stretch considering the source) will come in around 270. The only reason to buy a Phenom is if you have an AM2 board you want to upgrade.

either way now ther is a choice for people. atleast for mid to low which is all i want.
when a cpu first comes out of course it will be high. consider how long the Q6600, people just said it performs around a QX6850.

[Syn.]

“alf7_munif” gets brownie points for finding AMD Phenom specs online and posting the information in the HardForum. Looks like a 2.4GHz part for 299.90EUR or US$436.60.

http://www.hardocp.com/news.html?new...VzaWFzdCwsLDE=

[ryboto]

They tell you in the article:

"measured before the VRM"


Meaning they actually strapped a potentiometer onto the soldered leads at the motherboard as it was running (or more likely, before they hit the power button).

This is by far the most accurate method for calculating power consumption.

Well, I don't trust it unless they explain it to me, or show me. There are a lot of soldered leads all over the VRM circuitry, and an explanation of "measured before the VRM" isn't all that descriptive or scientific.

[ryboto]

For this one:
http://www.matbe.com/images/biblio/a...0000063320.png

They measure current deliver to the cpu with something called a "pince ampérométrique." I believe translation is "clamp-on ammeter"

http://en.wikipedia.org/wiki/Clamp_meter

ok...that's still confusing, how can you fit a clamp ammeter on a motherboard? Or are they measuring current given to the board?

[informal]

Well, I don't trust it unless they explain it to me, or show me. There are a lot of soldered leads all over the VRM circuitry, and an explanation of "measured before the VRM" isn't all that descriptive or scientific.

Yep ,this is very true.

[nemrod]

ok...that's still confusing, how can you fit a clamp ammeter on a motherboard? Or are they measuring current given to the board?

They measure current on the atx cable, so before the vrm.

[ryboto]

They measure current on the atx cable, so before the vrm.

so, the atx cable is the only 12v line the CPU gets it's power from?

[nemrod]

so, the atx cable is the only 12v line the CPU gets it's power from?

As far I understand, yes

[Syn.]

guys this is a "Pricelist" thread, take your Voltage talks to a more suitable thread.

[mstp2009]

so, the atx cable is the only 12v line the CPU gets it's power from?

Why don't you email them? I'm sure they would be happy to explain their testing methods further.

[ryboto]

Why don't you email them? I'm sure they would be happy to explain their testing methods further.

I was just curious, but like others, it's baffling to me that Intel would throw such a high TDP rating if that really is the power draw. If true, Penryn mobile parts are going to be amazing. As for Phenom prices, I can't justify myself spending more than $200 on a computer component. Then again, if the performance/watt/price is greater than my 4000+, I might be tempted.

[red]

http://download.intel.com/design/Xeo...s/30235501.pdf

It is the highest expected sustainable power while running known power-intensive real
applications.

This means that if stretched towards certain limits, it can handle 130W. It is not "really the power draw" (aka average power use or even power use during good but not extreme loading). Capeesh?

[JumpingJack]

http://download.intel.com/design/Xeo...s/30235501.pdf

This means that if stretched towards certain limits, it can handle 130W. It is not "really the power draw" (aka average power use or even power use during good but not extreme loading). Capeesh?

Unfortunately, this is the prevailing mentality to TDP... i.e. people errantly equate TDP = consumption, and it is not. AMD even invented a new metric, ACP... frankly, an effective marketing tool to put a lower 'wattage' number next to the CPU, and it has fooled many people into mixing these two 'specs'.

All of Intel's TDP definitions have been 'worst case' power applications, yet AMD has effectively convinced people that it means 'average' power consumption.

The key is to simply measure it under load to determine the actual consumption.

[JumpingJack]

how can they measure at the socket?

You put a shunt at the +12VDC plug that powers the CPU and measure the voltage drop and current at that plug. Unfortunately, it also includes the inefficiencies of the VRM, which is why he makes the 70% efficiency translates into actual 50 Watt consumption statement.

A few other sites do this type of measurement (Xbitlabs, Digit-Life), here is a description of how they do it... there are a few different ways to getting there:

http://www.digit-life.com/articles2/...2-5-final.html
Skip to near the bottom where they discuss testing CPU power methodologies.

There are advantages and disadvantages to getting this type of data, first it takes out any ambiguity from the system and focuses more on the CPU, however as Intel has the mem controller on the chipset and AMD puts it on the CPU, this is not an apples to oranges comparision. However, Intel makes up for the mem BW degradation with larger caches, so Intel uses more transistors -- to the 'deficit' that AMD would have you believe is not as severe or more likely not at all when comparing ... but it is still apples to oranges.

What you can conclude from this is that per SOCKET the power consumed and dissipated.... which is important for people looking to thermal solutions.

In short, if you are interested in seeing what it may do to your electric bill -- system power is what you want, if you are interested in picking out your thermal solution then socket power is what you want.

Jack