Thermal design power (TDP): Foundations — Hardware deep dive

This page assumes nothing. Before you can read the parent note Thermal design power (TDP), you need to know what a watt is, what heat even means, why a chip gets hot at all, and how engineers borrowed the language of electricity to describe temperature. We build every symbol from the ground up.

1. Power, energy, and the watt (the unit TDP is measured in)

Start with the two words people mix up constantly.

Figure s01 — The tap analogy. A tap pours water into a puddle. The thickness of the falling stream (blue) stands for power — how fast energy flows, in watts. The puddle piling up on the floor (yellow) stands for energy — the total amount collected, in joules. Same tap, two different quantities: rate vs. total.

Look at the tap picture. The flow rate (thickness of the stream) is power; the puddle that piles up is energy. TDP is a power number, because a cooler cares about the rate heat arrives, not the total over a whole day.

2. Heat and temperature (two different things)

We will meet three temperatures later. Give them names now:

$T_{j}$ — junction temperature, the temperature of the actual silicon inside the chip. "Junction" is old transistor slang for the working guts of the chip.
$T_{ambient}$ — the temperature of the surrounding air the cooler dumps heat into.
$T_{j,max}$ — the highest junction temperature allowed before the chip protects itself. Cross it and the chip slows down (see Thermal throttling) or dies.

3. Electric charge, current, and capacitance (the electrical building blocks)

Before we can say why a chip heats up, we need three electrical words. Build them in order.

Now every electrical symbol we need — $Q$ , $I$ , $V$ , $C$ — has a plain meaning, a unit, and a picture. We are ready to explain heat.

4. Why a chip produces heat at all

Every chip is built from millions of tiny switches called transistors. When a transistor flips on and off, it moves electric charge around. Moving charge through the chip's material meets resistance, and — exactly like a light bulb filament — that turns electrical energy into heat.

Where $P_{dyn} = α C V^{2} f$ comes from

Before we accept the formula, let us build it, one switching event at a time.

The physics of charging a capacitor of size $C$ up to voltage $V$ gives an energy stored of $\frac{1}{2} C V^{2}$ . But charging it from the supply pulls a total of $C V^{2}$ out of the power source (half is stored, half is lost as heat in the wires on the way in), and then emptying it dumps the stored half as heat too. Over one full charge-then-discharge cycle the energy that ends up as heat is:

$E_{sw} = C V^{2}$

Recall What happened to the textbook "½" in

$\frac{1}{2} C V^{2}$ ? $\frac{1}{2} C V^{2}$ is the energy stored in the capacitor ::: but over a complete charge and discharge cycle both halves eventually become heat, giving a full $C V^{2}$ per cycle — so no ½ survives in $P_{dyn}$ . (Some textbooks fold the factor into $C$ or $α$ ; conventions vary, so check which energy your source means.)

The total the cooler must handle: $P_{total} = P_{dyn} + P_{static}$

Recall Why does

$V$ appear squared in dynamic power? Because voltage does double duty ::: it sets both how much charge is moved (the bucket fills to $Q = C V$ ) AND how hard it is pushed, so its effect multiplies with itself, giving $V^{2}$ .

5. The Ohm's-Law analogy for heat (the heart of every TDP calculation)

Here is the clever borrowing that makes TDP math easy.

In electricity, Ohm's Law says a voltage difference $Δ V$ pushes current $I$ (charge per second, in amperes — defined in §3) through a resistance $R$ : $Δ V = I \times R$

Engineers noticed heat behaves the same way: a temperature difference pushes heat flow through a thermal resistance. So they copied the equation symbol-for-symbol.

Figure s02 — The electrical–thermal analogy. Top (blue): a voltage difference from "high V" to "low V" drives a current $I$ through a resistor $R$ . Bottom (pink): a temperature difference from the hot junction $T_{j}$ to the cool air $T_{ambient}$ drives a heat flow $P$ (watts) through a thermal resistance $R_{θ}$ . The two rows are the same equation with the labels swapped.

Formula The thermal Ohm's Law

$Δ T = P \times R_{θ}$ Defining the temperature difference for a chip. The general form is $Δ T = T_{hot} - T_{cold}$ , where $T_{hot}$ is simply "the temperature at the hotter end of the heat path" and $T_{cold}$ is "the temperature at the cooler end". For a CPU the hot end is the silicon junction and the cold end is the surrounding air, so we substitute the chip-specific names from §2: $T_{hot} = T_{j}, T_{cold} = T_{ambient}, Δ T = T_{j} - T_{ambient}$ Taken as (hotter end) minus (cooler end), $Δ T > 0$ , and heat always flows from hot to cold.

$Δ T$ — temperature difference (°C), plays the role of voltage.
$P$ — heat flow in watts, plays the role of current $I$ .
$R_{θ}$ — thermal resistance ("R-theta"), how much the path resists heat flow, measured in °C/W. Since it is a difference of temperatures, one degree Celsius equals one kelvin, so °C/W and K/W are numerically identical — you will see both symbols; treat them as the same unit.

The little $θ$ (Greek "theta") is just a subscript label meaning "thermal" — it tells you this R is about heat, not electricity.

Because the heat travels a chain of stages, and resistances in a chain simply add up (just like resistors in series), the parent note writes: $R_{θ, total} = R_{θ, JC} + R_{θ, CS} + R_{θ, SA}$

Figure s03 — Heat crossing three resistances in series. Boxes left to right: silicon junction ( $T_{j}$ ) → metal case → heatsink fins → ambient air ( $T_{ambient}$ ). The three arrows between them are labelled $R_{J C}$ (junction-to-case), $R_{C S}$ (case-to-sink, the thermal paste), and $R_{S A}$ (sink-to-ambient). Heat is a hiker crossing all three bridges in a row, so the resistances add: $R_{total} = R_{J C} + R_{C S} + R_{S A}$ .

Read the picture left to right — heat is a hiker crossing three bridges:

$R_{θ, JC}$ — Junction-to-Case: silicon out to the metal lid.
$R_{θ, CS}$ — Case-to-Sink: across the thermal paste (see Heatsink design and thermal resistance).
$R_{θ, SA}$ — Sink-to-Ambient: from the heatsink fins into the moving air.

6. Steady state vs. spikes — thermal capacitance (why TDP is a sustained number)

The thermal Ohm's Law above is a steady-state picture: it assumes heat has been flowing long enough that temperatures have settled. But chips also have thermal mass, and that changes how they respond to short bursts.

Figure s04 — Junction temperature over time after a sudden load. Yellow curve: apply constant power and $T_{j}$ rises gradually (not instantly) toward its steady value, because the thermal capacitance must fill up first. Blue dashed line: the final steady-state temperature the thermal Ohm's Law predicts. Pink arrow: a brief power spike only climbs partway up the curve before the load drops — which is exactly why a chip can momentarily exceed TDP without overheating.

7. Putting the symbols together

First, the missing link between the two halves of this page: TDP is just a name for the chip's steady-state total heat flow.

Now rearrange the thermal Ohm's Law to see where TDP fits. The chip warms until heat removed equals heat made. At that balance the heat flow $P$ equals TDP, so: $TDP = \frac{T _{j} - T _{ambient}}{R _{θ, total}}$

Every symbol on the right you now know. To keep the chip safe we demand $T_{j} \leq T_{j,max}$ , which rearranges to the cooling requirement: $R_{θ, total} \leq \frac{T _{j,max} - T _{ambient}}{TDP}$

Recall If

$T_{ambient}$ rises on a hot day, what happens to your allowed $R_{θ, total}$ ? The numerator $T_{j,max} - T_{ambient}$ shrinks ::: so the allowed thermal resistance drops — the same cooler may no longer be good enough, which is why servers care about Data center cooling.

Prerequisite map

The diagram below is a dependency map: read it top to bottom, following the arrows. Each box is one idea from this page, and an arrow $X \to Y$ means "you need $X$ before $Y$ makes sense" — $X$ feeds into $Y$ . The two independent roots (energy-in-joules on the left, the electrical Ohm law / $P = V I$ on the right) flow downward and both meet near the bottom at the "cooling requirement", which is the formula the parent note is built on. Figure s05 draws the same map as a labelled picture in case the diagram below does not render.

Figure s05 — The prerequisite map drawn as a picture. Yellow nodes on the left are the energy/heat chain; blue nodes on the right are the electrical chain ( $P = V I$ , charge, current, capacitance); pink nodes at the bottom are where both chains merge into the thermal Ohm's Law and the final cooling requirement. Arrows point from each prerequisite to what it enables.

Equipment checklist

Test yourself — if you can answer each, you are ready for the parent note.

What unit is TDP measured in, and is it a rate or a total?

Watts (W); it is a rate of heat flow (joules per second), not a total amount of energy.

What is the difference between heat and temperature?

Heat is energy in transit (measured in joules, or watts as a flow rate); temperature is "how hot" something is — a level, in °C.

What are electric charge and current, and their units?

Charge is a quantity of electricity in coulombs; current is the rate charge flows, in amperes (1 A = 1 coulomb/s).

State the master power relation and its units.

P = V \times I

— voltage (volts) times current (amperes) gives power (watts).

What is capacitance, its unit, and its defining equation?

How much charge a bucket holds per volt; measured in farads;

Q = C V

Why does essentially all electrical power in a CPU turn into heat?

The chip does no mechanical work, so by conservation of energy every joule drawn must leave as heat.

How does

P_{dyn} = α C V^{2} f

follow from energy per switch?

Each full switch dumps

E_{sw} = C V^{2}

of heat; multiply by the switching rate

α f

(fraction switching times ticks per second) to get power.

Where does

P_{static} = I_{leak} V

come from?

It is just the master relation

P = V I

applied to the leakage current

I_{leak}

State the thermal Ohm's Law and its sign convention.

Δ T = P \times R_{θ}

with

Δ T = T_{hot} - T_{cold} = T_{j} - T_{ambient}

, so heat flows from the hotter end to the cooler end.

What are the units of thermal resistance

R_{θ}

, and does low or high mean a better cooler?

°C/W, identical to K/W for temperature differences; low is better — little temperature rise per watt.

How is TDP related to

P_{total}

, and why is it not the absolute worst case?

TDP is

P_{total}

at steady state under the maker's chosen benchmark workload; stress tests, overclocks, and turbo bursts can exceed it.

What is thermal capacitance and why does it let a chip briefly exceed TDP?

The heat the mass must absorb to rise 1 °C; it delays the temperature rise, so short spikes only climb partway up the curve before the load drops.

Why do

R_{θ, JC}

R_{θ, CS}

R_{θ, SA}

add together?

Heat flows through them in series like resistors in a chain, and series resistances add.

What do

T_{j}

T_{ambient}

, and

T_{j,max}

each refer to?

Silicon junction temperature, surrounding air temperature, and the maximum junction temperature allowed before throttling.