What Is TDP (Thermal Design Power)? CPU and GPU Explained

TDP, or Thermal Design Power, is the amount of heat (measured in watts) that a chipmaker expects a CPU or GPU to produce during sustained, real-world use — and therefore the minimum amount of heat its cooler needs to be able to remove. It is a cooling-design spec, not a direct measurement of how much electricity the chip actually pulls from the wall, which is why a "125W" processor can legitimately draw well over 200W when it's boosting. Understanding that distinction — and the related PL1/PL2/PPT numbers Intel and AMD publish alongside it — is the key to picking a cooler and power supply that won't bottleneck or underpower a new build.

What Is TDP (Thermal Design Power)?

Thermal Design Power describes the maximum amount of heat a CPU, GPU, or system-on-chip is expected to generate while running real applications at its base clock speed, expressed in watts. According to Wikipedia's entry on thermal design power, it tells system builders and cooler manufacturers how much heat-dissipation capacity a cooling solution needs — "the maximum amount of heat generated by a cooling device that is required to dissipate during normal operation." It originated as a cooling-engineering number, not a consumer power-draw spec, which is why it's always expressed in watts of heat rather than watts of guaranteed electrical draw.

Crucially, the same Wikipedia summary notes that TDP values "cannot be accurately compared" across manufacturers, because each vendor defines its own safety margin and its own idea of a representative workload. A 65W AMD rating and a 65W Intel rating are not measured the same way, and the methodology has shifted over time — AMD historically reported close to peak power draw as TDP, while Intel has used a more conservative, sustained-workload figure since its Core 2 "Conroe" generation. That history is part of why the same wattage number can mean different things depending on whose sticker it's printed on.

TDP vs. Actual Power Draw: Why They're Not the Same

TDP is a design target, not a live power meter reading. A chip's real power draw moves constantly depending on workload, temperature, silicon quality, and how aggressively its motherboard or BIOS lets it boost. That's true on both CPUs and GPUs. Nvidia's own engineering writeup on RTX 40-series efficiency is explicit about this gap: it describes Total Graphics Power (TGP) as "the power cap limit" a card is allowed to hit — not what it typically draws. In Nvidia's own five-game test data, an RTX 4080 rated at a 320W TGP averaged just 251W in actual gameplay, and never touched its full rated ceiling even at 4K with ray tracing enabled.

CPUs behave the same way, just in the other direction: short-term draw can spike well above the base TDP number. A chip rated for a 125W base TDP might briefly pull 250W or more while boosting on a handful of cores, provided the cooler can keep temperatures in check and the motherboard's power limits allow it. This is exactly what Intel's PL1/PL2 system and AMD's PPT system exist to control — they define how high and how long a chip is allowed to exceed its base rating before it's forced back down.

Intel's Power Limits: PL1, PL2, PL3, PL4, and Tau

Since 12th-generation Core processors, Intel has formally retired the term "TDP" for desktop and mobile chips in favor of two numbers: Processor Base Power (PBP) and Maximum Turbo Power (MTP). Underneath those consumer-facing labels sits the same power-limit framework Intel has used for years, documented in its processor datasheets as PL1 through PL4:

• PL1 — "A threshold for average power that will not exceed," which Intel recommends setting equal to Processor Base Power. This is the sustained, long-run limit — effectively the modern name for base TDP.
• PL2 — "A threshold that if exceeded, the PL2 rapid power limiting algorithms will attempt to limit" further spikes. This is the short-term turbo ceiling, and it typically lines up with the published Maximum Turbo Power figure.
• PL3 — A faster-reacting limit that curbs the duty cycle of brief power spikes above PL3 by reactively trimming frequency.
• PL4 — A hard ceiling the chip is never allowed to exceed; PL4 algorithms preemptively cut frequency to prevent any spike from crossing it.
• Tau (Turbo Time Parameter) — the averaging window used to calculate PL1 as a rolling average, rather than an instant reading.

These definitions come straight from Intel's Package Power Control documentation. In practice, a chip can legally run at its PL2 wattage for the length of its Tau window — often 28 to 56 seconds on desktop parts — before it's averaged back down toward PL1. On the Core Ultra 9 285K, for instance, Intel's own specification page lists a 125W Processor Base Power alongside a 250W Maximum Turbo Power — meaning the chip is designed to sustain 125W of heat output indefinitely, but can legitimately pull roughly double that during bursts of heavy, multi-core work.

AMD's Power Limits: TDP, PPT, TDC, and EDC

AMD still uses the term TDP directly on its Ryzen product pages, but — like Intel's PBP/MTP pair — it's only part of the picture. Ryzen desktop chips are governed by three limits enforced through the motherboard firmware and AMD's own tuning software:

• PPT (Package Power Tracking) — the socket-level power ceiling, in watts, set higher than the rated TDP to give the chip headroom for short all-core boosts.
• TDC (Thermal Design Current) — the sustained current limit, in amps, tied to what the motherboard's voltage regulators can supply continuously.
• EDC (Electrical Design Current) — a higher, short-duration current limit for brief spikes.

Precision Boost Overdrive (PBO) — AMD's equivalent of the overclocking headroom Intel exposes through PL2 — works by raising or removing these PPT/TDC/EDC ceilings so the chip can sustain higher clocks for longer, at the cost of higher power draw and heat. If you're weighing whether PBO or manual tuning is worth it on your own build, our plain-English guide to overclocking CPUs, GPUs, and RAM walks through what each setting is actually doing. On current AM5 desktop chips, published TDP ratings range from 65W on efficiency-focused parts up to 170W on flagship multi-core parts — full figures are in the comparison table below.

GPU Power Ratings: TGP, TBP, and Why the Term Changes

Graphics cards almost never carry a "TDP" badge the way CPUs do. Nvidia's reference and Founders Edition specification pages list a Total Graphics Power (TGP) figure instead — for example, Nvidia's RTX 5090 specifications page lists a 575W TGP for the Founders Edition design, while the RTX 5080 and outgoing RTX 4090 are rated at 360W and 450W respectively on Nvidia's own card pages (see the comparison table below for how that stacks up against a current CPU). TGP covers the whole board — GPU die, memory, and onboard power circuitry — which is why it tends to run noticeably higher than a CPU's base TDP for a comparable die size. AMD and board partners often use the near-identical term TBP (Total Board Power) for the same concept. Whichever label is used, treat it the same way you'd treat a CPU's turbo figure: it's the power ceiling the card's boost algorithm is allowed to hit, not a guarantee of what it will draw in every game.

Why TDP Matters for Cooling Your PC

A cooler has to remove heat at least as fast as the chip produces it, or the processor will throttle its clock speed down to stay within its thermal limits — quietly costing you performance you paid for. Base TDP/PBP tells you the floor a cooler needs to clear for sustained workloads; the turbo figure (PL2/MTP/PPT) tells you what it needs to handle during bursts without the chip immediately backing off. A cooler rated right at a chip's base TDP will usually keep temperatures in check for lighter, everyday tasks, but it can struggle the moment a sustained multi-core workload — video encoding, compiling, heavy gaming — pushes the chip toward its turbo ceiling for more than a few seconds. If you're choosing between an all-in-one liquid cooler and a high-end air cooler for a higher-TDP chip, our AIO vs. air cooler comparison breaks down which setups actually justify the extra cost. As a rule of thumb, buy cooling capacity for the turbo number on the spec sheet, not the quieter base number.

Why TDP Matters for Power Supply Sizing

The same logic applies to your power supply, with one extra wrinkle: PSUs also have to absorb brief, millisecond-scale transient spikes that can run well above even the turbo/TGP rating, especially on high-end GPUs. A practical sizing approach is to add up the CPU's turbo power figure (MTP or PPT) and the GPU's TGP/TBP figure, then build in headroom — many enthusiasts and system integrators target roughly 30% extra capacity above that combined peak to comfortably absorb spikes and leave the PSU running in its efficient mid-load range rather than near its ceiling. Undersizing a PSU around the lower base-TDP number is one of the more common build mistakes, and it tends to show up as random shutdowns under load rather than an obvious error message.

TDP and Power Ratings Compared: Current CPUs and GPUs

The table below lines up published base and turbo/boost power figures for current flagship-to-mainstream CPUs and GPUs, straight from each maker's own specification pages. Note how differently "TDP" behaves across the two CPU vendors and how much higher GPU TGP figures run by comparison.

Intel figures are from Intel's product specification pages; AMD Ryzen 9000-series figures are drawn from publicly documented specifications compiled on Wikipedia's Zen 5 architecture page, since AMD's own product pages could not be reached at the time of writing; Nvidia figures are from each card's official specification page on nvidia.com. GPU rows have no separate "base" figure because TGP is itself a single boost-level ceiling, not a base/turbo pair.

What to Do With This Number When Building a PC

In practice, TDP and its siblings are most useful as a floor, not a target. When shopping for a cooler, match or exceed the chip's turbo-level rating (MTP for Intel, or a healthy margin over the base TDP for AMD) rather than its quieter base figure. When shopping for a case, make sure its airflow is rated for roughly the combined turbo output of your CPU and GPU, since both are dumping heat into the same enclosure. And when comparing two chips on paper, remember the warning baked into the definition itself: a 65W rating from one vendor and a 65W rating from another are not measured the same way, so TDP alone is a poor substitute for an actual buying decision. Our broader 2026 CPU buying guide comparing Intel and AMD covers how to weigh power figures against real performance and pricing.

What to Read Next

TDP is one piece of a larger cooling-and-power picture. If you're planning an upgrade, pair this explainer with a look at how much performance headroom overclocking or power-limit tuning can actually unlock, and make sure whatever cooler or PSU you land on is sized for the turbo number — not the quieter base rating on the box.

Written byMara Lindqvist

Mara Lindqvist edits the hardware desk. She covers graphics cards, processors, memory and storage, the foundries and chip designers behind them, and what the numbers on a spec sheet mean for people choosing a PC. Specifications in her stories come from manufacturer spec pages and datasheets.

More from Mara LindqvistAll Hardware coverage