Wall Power Measurement: What Your PC Actually Draws vs Component Power Ratings

Adding up a CPU's rated TDP and a GPU's rated board power to estimate what a system draws from the wall consistently undercounts, because that arithmetic ignores PSU conversion losses and the fact that rated figures rarely describe worst-case combined load.

Plugging a system into an inline power meter at the wall outlet and running a combined CPU and GPU stress test typically produces a number noticeably higher than simply adding the CPU's rated power figure to the GPU's rated board power figure. The gap comes from two separate sources: PSU conversion inefficiency, and the fact that component-rated power figures describe a specific test condition that a real combined workload does not always match.

Where the Gap Comes From

The PSU efficiency loss alone typically accounts for most of the gap. An 80 PLUS Gold PSU is only guaranteed to hit around 87 to 90 percent efficiency at specific load points defined by the certification test, and efficiency drops off measurably at both very low load (well under 20 percent of rated capacity, common on an idle desktop) and very high load (above 90 percent of rated capacity). A system that draws 400W of DC power to its components might pull 460 to 480W from the wall, not 400W, because that DC power had to be converted from AC with some loss along the way.

Why This Matters Beyond Curiosity

Beyond satisfying curiosity about an electricity bill, wall power measurement is the only reliable way to catch an undersized PSU before it causes instability under real combined load, since manufacturer TDP and board power figures for the CPU and GPU individually do not account for how those parts behave when both are under heavy load simultaneously, nor do they account for transient spikes that a cheap or undersized PSU may not ride through cleanly.

• Use a meter with a fast enough sample rate to catch spikes, not just an average reading over several seconds; some budget meters smooth readings in a way that hides brief transient spikes entirely.
• Test under the heaviest realistic combined load your system will see (a demanding game, not just a synthetic single-component stress test), since real workloads load CPU and GPU simultaneously in ways synthetic single-component tests may not.
• Compare the measured wall draw against your PSU's rated capacity with headroom to spare, not right up against the limit, since PSU efficiency and component power draw both vary with temperature, age, and silicon variance between otherwise identical units.

A wall meter reading that comes in significantly higher than expected is also a useful early diagnostic for a failing PSU, since conversion efficiency degrades as electrolytic capacitors age, meaning the same system can measurably draw more from the wall for identical DC output several years into a PSU's service life.

Sizing a PSU Using Wall Power Data

Once you have a realistic wall power figure under peak combined load, sizing a PSU with adequate headroom is straightforward: aim for the PSU's rated capacity to sit comfortably above the peak wall draw, not the sum of component DC ratings, and ideally land near the PSU's most efficient load range rather than right at its ceiling. A PSU running consistently near 90 percent or more of its rated capacity spends more time in a less efficient part of its curve and has less margin to absorb the brief transient spikes that GPUs and CPUs are both known to produce under sudden load changes, even if sustained average draw looks comfortably within spec.

This is also why two builds with seemingly identical component lists can call for different PSU capacities in practice: a system with several extra drives, more case fans, or RGB lighting controllers draws meaningfully more baseline wattage than a minimal build with the same CPU and GPU, and that difference only shows up in a wall measurement, not in a simple CPU-plus-GPU spec sheet estimate.