What Is TDP? CPU Thermal Design Power Explained

TDP (thermal design power) is the average heat, in watts, that a processor’s cooling system must be able to remove while the chip runs at its base frequency under a demanding workload. It is a cooling target set by the chip maker, not a measure of speed and not the most power the processor can draw.

That is the short answer to what TDP means, and to what TDP stands for. The figure tells a system designer how big the heatsink, heat pipes or fan need to be. Intel has called the same figure processor base power since its 12th Gen Core processors, and both Intel and AMD let a processor run above it for short periods when there is thermal headroom.

For an industrial PC the number matters more than it does on a desk. A sealed, fanless chassis has a fixed ability to shed heat, so the processor’s TDP decides how hot the enclosure can get, how much of the turbo performance you keep, and how long the system will run without slowing down. This guide explains TDP alongside the power limits that now sit around it (PL1, PL2 and cTDP), then shows how to size a fanless computer around them.

Processor datasheets now list several power figures. They describe different things, and mixing them up is the most common reason a system runs hotter or slower than expected.

Two corrections to older guides, including an earlier version of this one. First, maximum turbo power is Intel’s current name for the PL2 value, not for cTDP. Second, TDP is not a power consumption figure: during turbo the processor can draw well above it for a while, and at idle it draws far less.

How PL1 and PL2 work together

When a heavy task starts, the processor is allowed to rise to PL2 and run at high turbo frequencies. It tracks its average power over the Tau window, and once that average reaches PL1 it drops back to the sustained limit. If the cooler cannot hold the die below TjMax at either level, the processor slows itself further; our guide to thermal throttling explains what that looks like in practice. On desktop motherboards PL2 is sometimes left on indefinitely, which is why the same processor can score very differently from one system to the next.

AMD and GPU equivalents

AMD lists a TDP for its processors too, and on socketed desktop Ryzen it also sets a package power tracking (PPT) limit above the TDP that plays a similar role to PL2. AMD embedded and mobile processors usually quote a configurable TDP range rather than one number.

Graphics processors follow the same idea under different names. Discrete GPUs list total graphics power or total board power, and MXM modules usually have a range the system maker chooses within. NVIDIA Jetson modules use selectable power modes, from about 7 W on the smallest Orin Nano modules to 60 W on Jetson AGX Orin 64GB. Our Jetson thermal design guide covers how those modes map onto a cooling design.

Manufacturer ratings show the trade-off clearly. In the same fanless chassis, moving from a 35 W to a 65 W processor typically takes 15–20 °C off the top of the operating range, while the cold end stays the same.

The Nuvo-9531-FT shows the other route: rather than accept a narrower temperature range, it runs 65 W desktop processors at a 35 W sustained limit, keeping most of their core count while staying inside a fanless envelope. The SPC-7000 does the same at the low end with configurable TDP on 15 W mobile processors.

A fanless design moves heat from the processor through a heat spreader and heat pipes into the finned case. Our guides to how a fanless PC works and to active vs passive cooling go further, and sealed or rugged systems can also use conduction cooling into a cold plate or vehicle chassis.

Embedded processors cluster into three TDP classes, and most industrial projects can be placed in one of them before looking at model numbers.

In Intel’s embedded line the suffix tells you the class: “E” marks an embedded part with long-term availability, and “TE” marks the 35 W version. Our guide to embedded processors explains the naming in more depth, and Intel Core Ultra covers the newer processors that add an NPU inside a similar power budget.

Does a higher TDP mean better performance?

Within one generation, a higher TDP usually buys more sustained performance, because the processor can hold higher clocks or run more cores before it reaches its limit. Across generations it does not: a newer 35 W processor can outperform an older 65 W one. Compare benchmark results for the workload you plan to run, then pick the lowest TDP class that meets it. Hybrid processors complicate the picture further, as P-core vs E-core sets out.

TDP and running costs

TDP also gives an upper bound on energy use. A processor running at its rated power around the clock uses its TDP in watts multiplied by 8,760 hours a year:

Real systems spend much of their time below base power, so treat these as worst-case figures and multiply by your own tariff. The gap between classes is still useful: across a 100-system deployment, choosing 35 W over 65 W processors saves up to about 26,000 kWh a year, and the lower heat usually means a smaller, sealed chassis.