PARKE wrote on 2022-06-21, 13:13:

AlexZ wrote on 2022-06-21, 11:22:

Undervolting the CPU will make it draw less current. Instead of 1.75V, use 1.7V.

Are you sure ?

I can confirm this with measurements I have taken. You may be thinking of inductive loads. A cpu basically acts like a resistive load. Though as others have pointed out they contain capacitive elements. Which are different yes, but less voltage is also less load for them. The load will decrease with lower voltage. As my measurements have shown. Amps will fall at the same time as volts and of course, maximum clock speed will as well.

Oetker wrote on 2022-06-21, 15:54:

appiah4 wrote on 2022-06-21, 11:27:

rmay635703 wrote on 2022-06-20, 11:29:

With folks using a slocket to install Tualatins to 440bx boards why wouldn’t similar magic work here?

The BIOS probably doesn't have the microcode for a Tualatin let alone a late Coppermine. Moreover the board may not even have a VRM capable of going low enough to not fry a Tualatin.

A VRM good enough for Coppermine is good enough for Tualatin. Microcode isn't really an issue either (440BX boards that work with Tualatin don't have it (unless patched)), it really is a matter of how strict the BIOS is.

Not necessarily? For coppermine any VRM that can go as low as 1.7V is fine, but for Tualatin you want one that can go as low as 1.4V - 1.7V can be quite high for Tualatin and the board may just flat out refuse to POST because it can't recognize the required Vcore?

Sphere478 wrote on 2022-06-21, 22:01:

PARKE wrote on 2022-06-21, 13:13:

AlexZ wrote on 2022-06-21, 11:22:

Undervolting the CPU will make it draw less current. Instead of 1.75V, use 1.7V.

Are you sure ?

I can confirm this with measurements I have taken. You may be thinking of inductive loads. A cpu basically acts like a resistive load. Though as others have pointed out they contain capacitive elements. Which are different yes, but less voltage is also less load for them. The load will decrease with lower voltage. As my measurements have shown. Amps will fall at the same time as volts and of course, maximum clock speed will as well.

Yes, clockspeed will also increase current . Take the same simple capacitance equation and solve for capacitative current:

Icap = (dV / dt) * C

So the faster your CPU is, the more you refresh the capacitors during a given t period; as a result the current drawn over that certain period of time invariably increases.

appiah4 wrote on 2022-06-22, 06:00:

Oetker wrote on 2022-06-21, 15:54:

appiah4 wrote on 2022-06-21, 11:27:

The BIOS probably doesn't have the microcode for a Tualatin let alone a late Coppermine. Moreover the board may not even have a VRM capable of going low enough to not fry a Tualatin.

A VRM good enough for Coppermine is good enough for Tualatin. Microcode isn't really an issue either (440BX boards that work with Tualatin don't have it (unless patched)), it really is a matter of how strict the BIOS is.

Not necessarily? For coppermine any VRM that can go as low as 1.7V is fine, but for Tualatin you want one that can go as low as 1.4V - 1.7V can be quite high for Tualatin and the board may just flat out refuse to POST because it can't recognize the required Vcore?

That's true, but in practice a VRM's suitable for Coppermine (VRM spec 8.4) go down to 1.3v. Unless there's VRMs out there that only partially implement the spec, of course.
And as mentioned somewhere else in the thread, identical TDP at a lower voltage means the MOSFETs and board traces will need to deal with more current.