s939 TDP jump?

what the heck? jumps form 80 to 104 watts. are the s939 chips on 90nm now like prescott?

I just think that they used a wierd way to "test" those TDP values and i wouldn't trust them at all.

Yes the 90nm chips are going to have similar heat to the Prescott, SOI is a better alternative to strained silicon, but the problem isn't with intel's process, it is basic to 90nm and the signal loss. There was a front page article on this, let me see if I can find it.

http://www.overclockers.com/articles1025/

There ya go

Ir0nman

I was expecting that. Well....it seems like my next PC will be BTX using Intel's Conroe

Ir0nman said:

Yes the 90nm chips are going to have similar heat to the Prescott, SOI is a better alternative to strained silicon, but the problem isn't with intel's process, it is basic to 90nm and the signal loss. There was a front page article on this, let me see if I can find it.

http://www.overclockers.com/articles1025/

There ya go

Ir0nman

Click to expand...

x86-secret wrote that these 104W TDP ratings were found for current Socket 939 chips, which are 130nm. I wouldn't believe that x86-secret would get their hands on 90nm A64's before 130nm Socket 939 CPU's.

As i have said all along, if IBM and Intel are having issues with 90nm what made you think AMD with IBM fabs would be any different.

Be happy IBM has the yeilds up or we would be paying a lot more for a fx or opteron.

Well Intel does not have problems with 90 nm....Prescott have 3 times as many transistors if you compare to nortwood. So i don't think that going to 90nm will increse power consumption if you compare the 90nm part with an 130nm one....

Remember the AMD 90nm shrink is a clean shrink they aren't going to triple amount of trannies like Intel did.

i'd like to know where they got the numbers from, cuz they certainly aren't in AMD's white papers

x86-secret is (as usuall) wrong. I have seen non-public papers and can tell that TDP will stay the same. It will change with 90nm.

AMD 130nm: 60A * 1.5V = 90W
AMD 90nm: 80A * 1.3V = 104W

Intel 90nm (FMB 1.0): 70A * VID=1.4V = 100W
Intel 90nm (FMB 1.5): 91A * VID=1.4V = 115W

Intel CPUs use VRDown design, their maximum voltage supplied decreases with increasing Amps. Prescott FMB 1.5 has under full load a voltage of aproximately 1.26V.


Intel datasheets give TDP that is typical power while executing publicly available software. TDP is not equal to Amps * Vcc therefore it is not maximum power CPU can dissipate under worst conditions. AMD give TDP as a maximum power under worst conditions. You can't simply compare Prescott's TDP to A64 TDP because those numbers are each telling different fact.