#### Do you have a question? Post it now! No Registration Necessary. Now with pictures!

**posted on**

July 29, 2005, 9:11 pm

anything. Can someone remind me of the formula to calculate the thermal heat

output of a CPU given its voltage, frequency etc. I know I had it writte

down somewhere, but I can't remember where I put it.

It involves squaring the voltage and multiplying by the frequency and some

other bits, but the order and extra bits in the formula escape me...

Thanks.

## Re: My memory is useless...

On Fri, 29 Jul 2005 18:00:36 +0100, "GT"

That's because you probably used the wrong start-page :-)

I wrote: intel.com. Not google.com

www.intel.com/support/processors/sb/cs-011032.htm

www.intel.com/design/pca/applicationsprocessors/applnots/28001201.pdf

www.intel.com/design/intarch/applnots/24157503.pdf

www.intel.com/technology/magazine/computing/ac04002.pdf

And there's plenty more where this came from :-)

Happy reading.

--

Kind regards,

Gerard Bok

That's because you probably used the wrong start-page :-)

I wrote: intel.com. Not google.com

www.intel.com/support/processors/sb/cs-011032.htm

www.intel.com/design/pca/applicationsprocessors/applnots/28001201.pdf

www.intel.com/design/intarch/applnots/24157503.pdf

www.intel.com/technology/magazine/computing/ac04002.pdf

And there's plenty more where this came from :-)

Happy reading.

--

Kind regards,

Gerard Bok

## Re: My memory is useless...

On Fri, 29 Jul 2005 16:11:12 +0100, "GT"

Yes, square the voltage an multiply the frequency. The part

you're leaving out is that each CPU core design is another

variable. You would backwards solve for that variable per

each CPU core. Even better would be to solve for several

models in the same family around the targets or average thm

all, but for the sake of simplicity (to get a near-enough

ballpark figure),

X = W/ [(Freq)*(V)²]

Probably the generic equation you wanted. Plugging some

numbers into that to determine the specific CPU core design

constant, let's consider a hypothetic CPU with the following

specs:

CPU "X" with 100W @ 3000 MHz & 1.5V, default spec.

Putting that into above equation we get,

X = 100 / [3000 * (1.5)²

X

X = 0.0148

So if that CPU were at 2.0V but same Freq. we take same

equation as above to find wattage (but only for that same

specific CPU core with the 0.0148 value).

X = W/ [(Freq)*(V)²]

0.0148=W / [3000 * (2)²]

W= 0.0148

Or if that CPU were at 1.5V default but o'c to 4GHz Freq,

0.0148 = W / [4000 * (1.5)²]

W = 0.0148

W = 133.2

(should be 133.33 actually but we encounter a margin of

error because that 0.0148 variable was rounded off, was more

like 0.0148148148148/etc). Similarly all the variables have

a margin of error, your core voltage won't be spot-on

1.5000V nor bus frequency exactly same, etc,

Yes, square the voltage an multiply the frequency. The part

you're leaving out is that each CPU core design is another

variable. You would backwards solve for that variable per

each CPU core. Even better would be to solve for several

models in the same family around the targets or average thm

all, but for the sake of simplicity (to get a near-enough

ballpark figure),

X = W/ [(Freq)*(V)²]

Probably the generic equation you wanted. Plugging some

numbers into that to determine the specific CPU core design

constant, let's consider a hypothetic CPU with the following

specs:

CPU "X" with 100W @ 3000 MHz & 1.5V, default spec.

Putting that into above equation we get,

X = 100 / [3000 * (1.5)²

X

*** 3000 ***(1.5)² = 100X = 0.0148

So if that CPU were at 2.0V but same Freq. we take same

equation as above to find wattage (but only for that same

specific CPU core with the 0.0148 value).

X = W/ [(Freq)*(V)²]

0.0148=W / [3000 * (2)²]

W= 0.0148

*** 3000 ***4Or if that CPU were at 1.5V default but o'c to 4GHz Freq,

0.0148 = W / [4000 * (1.5)²]

W = 0.0148

*** 4000 ***2.25W = 133.2

(should be 133.33 actually but we encounter a margin of

error because that 0.0148 variable was rounded off, was more

like 0.0148148148148/etc). Similarly all the variables have

a margin of error, your core voltage won't be spot-on

1.5000V nor bus frequency exactly same, etc,

#### Site Timeline

- » Wattage date for graphic cards. Would ati radeon all-in-wonder 8500 or 7500 work in 230W P...
- — Next thread in » Computer Hardware

- » Performance on Synology DS-101+
- — Previous thread in » Computer Hardware

- » hardware, UEFI, dual-boot, USBstik ?
- — Newest thread in » Computer Hardware

- » Anyone Using ESET NOD32??
- — The site's Newest Thread. Posted in » Anti-Virus Software