seronxTruth :toast:

But it doesn't help that

Cads Phenom II under heavy load pushes in 150 Watts when rated 125 Watts

and his Sandy Bridge with the iGPU turned off pushes in 65 watts when rated for 80~? Watts

It is under heavy load right?

Under Estimation and Over Estimation

In this case AMD is closer to the definition of TDP but it isn't the max TDP

Yes, it does help. The fact of the matter is that TDP has nothing to do with anything, but the required cooling. IF TDP = MAX power draw, we'd not be able to overclock, because although my SB will draw ~65w-70w using the CPU alone, it draws 125w, which is close to double, for 4.5 GHz. You could even say, that TDP is a meaningless number with todays technology...Intel's SB chips will make sure they do nto exceeed that 95w TDP, and throttle, even if running cool, unless you manipulate the BIOS. In that instance, it would be max power draw, but only then.

Under more harsh environments, CPU will run hotter, and will require that 95W of cooling at stock clocks, as the iGPU actually consumes very little power.

You realyl need to be careful how you state thing, or I'm afraid the good ol' "Boy who cried wolf" may apply to you here @ TPU.


Now here's my thing.

We got 95w SB chips pulling 125w @ 4.5 GHz. It's very common for power draw to be much less.

We got Bulldozer STOCK @ 125W. It's likely to use more.

So, OK, we got a silicon problem. Is that AMD's fault, or GLoFo's fault?

I am very excited by the Bulldozer design, but not about GLoFo's silicon. You wanna knwo about real perforamnce? Ignore the banches, and leaks, and examine the silicon. 32nm FM1 chips are available now.

So you can post all these benchmarks, or whatever...the fact remains that we know Bulldozer is going to require more cooling, from an "overclocking perspective"; it's also going to draw more power, and that alone will affect thier success. Just liek my SuperPi numbers, really, aren't that important, per se, there are those that care about such things. Likewise, there are those that like low-power, but hella fast chips.

In effect, you are posting and hyping up Bulldozer, but far too many of us here have been burnt by the hype in the past, so we won't go for it. It doesn't matter who the source is...people will remain skeptical.

Unless, of course, it's ME, or perhaps W1zz, posting those numbers. ;)



:laugh:

hekyNot all OEM prebuilts use IGP. Only the really low-cost ones do.

The majority do, only customized or high end builds don't.

cadavecaYes, it does help. The fact of the matter is that TDP has nothing to do with anything, but the required cooling. IF TDP = MAX power draw, we'd not be able to overclock, because although my SB will draw ~65w-70w using the CPU alone, it draws 125w, which is close to double, for 4.5 GHz. You could even say, that TDP is a meaningless number with todays technology...Intel's SB chips will make sure they do nto exceeed that 95w TDP, and throttle, even if running cool, unless you manipulate the BIOS. In that instance, it would be max power draw, but only then.

Under more harsh environments, CPU will run hotter, and will require that 95W of cooling at stock clocks, as the iGPU actually consumes very little power.

You realyl need to be careful how you state thing, or I'm afraid the good ol' "Boy who cried wolf" may apply to you here @ TPU.


Now here's my thing.

We got 95w SB chips pulling 125w @ 4.5 GHz. It's very common for power draw to be much less.

We got Bulldozer STOCK @ 125W. It's likely to use more.

So, OK, we got a silicon problem. Is that AMD's fault, or GLoFo's fault?

I am very excited by the Bulldozer design, but not about GLoFo's silicon. You wanna knwo about real perforamnce? Ignore the banches, and leaks, and examine the silicon. 32nm FM1 chips are available now.

So you can post all these benchmarks, or whatever...the fact remains that we know Bulldozer is going to require more cooling, from an "overclocking perspective"; it's also going to draw more power, and that alone will affect thier success. Just liek my SuperPi numbers, really, aren't that important, per se, there are those that care about such things. Likewise, there are those that like low-power, but hella fast chips.

I effect, you are posting and hyping up Bulldozer, but far too many of us here have been burnt by the hype in the past, so we won't go for it. It doesn't matter who the source is...people will remain skeptical.

Unless, of course, it's ME, or perhaps W1zz, posting those numbers. ;)



:laugh:

Cad,

How accurate is the "Powers Package" reading on Hwmonitor? Does it give an accurate interpretation of usage or is it just an estimate? Reason i ask because at 4.5Ghz it states my 2500K is only pulling 89.96W in IBT

And if the FX CPUs have a PCI-E Controller you know what is going to out perform in GPU performance

So you think pci-e controller can be integrated onto AM3 cpu? :wtf:
You're some kind of noob or smt? :banghead:

XoRSo you think pci-e controller can be integrated onto AM3 cpu? :wtf:
You're some kind of noob or smt? :banghead:

I think he's just been high on PCP for the past few days

PestilenceCad,

How accurate is the "Powers Package" reading on Hwmonitor? Does it give an accurate interpretation of usage or is it just an estimate? Reason i ask because at 4.5Ghz it states my 2500K is only pulling 89.96W in IBT

I don't use software for any real monitoring except temps. I have found AIDA, RealTemp, HW Monitor, and many others totally wrong on many occasions.

If I beleived HW Monitor, @ stock, on certain boards, my 2600k would be pulling 115w. But clamp meters around the 8-pin and other plugs, combined with a killawatt, tell me a far different story.

hekyNot all OEM prebuilts use IGP. Only the really low-cost ones do.

bulk of the market

XoRSo you think pci-e controller can be integrated onto AM3 cpu? :wtf:
You're some kind of noob or smt? :banghead:

AM3+

Llano has a 32x(4 x 8) PCI-E Controller
Only 2 of the 8 are for the second gpu(PCIE_X16_0)
1 is for the Southbridge and the other is for the Display Ports

That is on FM1 yes but the FM1 socket and AM3+ socket are relatively the same socket in size
-----------------------------------
Zambezi is on the 32nm fabrication same as Llano

It can have a 16x to 64x PCI-E Controller
(Not in block format)

Bulldozer can't have an on-die PCIe controller because AM3+ boards still use discrete northbridges + southbridges and are backwards compatible with AM3 K10 Phenom IIs. Since FM1 is a new socket, AMD was able to integrate the northbridge + PCIe into Llano and just use the FCH as the only major chip on the board.

Jstn7477Bulldozer can't have an on-die PCIe controller because AM3+ boards still use discrete northbridges + southbridges and are backwards compatible with AM3 K10 Phenom IIs. Since FM1 is a new socket, AMD was able to integrate the northbridge + PCIe into Llano and just use the FCH as the only major chip on the board.

I got confused with the PCI-e Controller for the South Bridge <-- the only integrated PCI-e Controller in Zambezi
(It's a lot bigger than the HT Links so it is easy to mistake it for something else)
(A-Link Express)

The Northbridge Controller in Zambezi/FX just has a faster links and lower latency with the discrete Northbridge over Deneb/Thuban/Phenom II CPUs

Gotta say I wonder how long AM3+ will last. They probably want to follow Intel and their own Llano's lead and increase integration in the CPU. Trinity will probably be just that move actually. They'll keep AM3+ as a sort of LGA 1366 high end workstation product I imagine.

swaayeGotta say I wonder how long AM3+ will last. They probably want to follow Intel and their own Llano's lead and increase integration in the CPU. Trinity will probably be just that move actually. They'll keep AM3+ as a sort of LGA 1366 high end workstation product I imagine.

With a reputation of supporting platforms for a long time they have painted themselves into a corner and thus they HAVE to support them for a long time.

[H]@RD5TUFFWith a reputation of supporting platforms for a long time they have painted themselves into a corner and thus they HAVE to support them for a long time.

They don't have to do anything and i remember reading AM3+ will be phased out by H2 2012 for a new socket for Bulldozer Enhanced.

Do you think Intel changes sockets just to piss off people? Ofcourse not. They do it for new features and tweaks.

PestilenceThey don't have to do anything and i remember reading AM3+ will be phased out by H2 2012 for a new socket for Bulldozer Enhanced.

Do you think Intel changes sockets just to piss off people? Ofcourse not. They do it for new features and tweaks.

No they don't HAVE to, but they really do after gaining the reputation for such. Intel changes sockets all the time and people QQ to no end about it. You really seem to have misunderstood what I said. My point is Intel changes sockets and is vilified for it, AMD keeps the same sockets, and is praised for it at the price of performance. They will need to chaznge sockets eventually and IMO they should have done so for bulldozer. I will be skipping it since it will be phased out during Q2 2012.

PestilenceThey don't have to do anything and i remember reading AM3+ will be phased out by H2 2012 for a new socket for Bulldozer Enhanced.

Do you think Intel changes sockets just to piss off people? Ofcourse not. They do it for new features and tweaks.

Not in all the cases
From socket 423 to 478 not sure, it should have always been 478
From 478 to 775 well, yes, but first chipsets were awful. 915... my god, awful stuff
From 775 to 1366 yes, for the IMC alone. big jump.
1366 to 1156 was like WTF? (not a transition really but just another option)
1156 to 1155 again, WTF? (don't know what to think on this one)

I want Amd to come back to an LGA socket

PestilenceI want Amd to come back to an LGA socket

Didn't know AMD did LGA, apart from servers...