A PC that switches off during a heavy GPU load often points to the power path. A fast change in graphics-card demand can expose weak PSU transient response, too little capacity, a poor connector fit or a protection circuit doing its job.

Quick Answer

Return the GPU to stock settings, inspect and reseat every graphics power plug, then test a known repeatable load. Check the card maker's PSU guidance against the power supply's exact model, age and connector layout. Do not keep forcing shutdowns while guessing.

Use the NVIDIA graphics-card selection to check candidate card specifications, and compare the wider market through popular GPU choices. Build the final power plan around the named card and full PC rather than a tier label such as mid-range or high-end.

Rated Power Is Only Part of the Story

A GPU power figure describes a sustained design target under defined conditions. Workloads change from one scene to the next, and the card can request power in short bursts as clocks and utilisation rise. A well-matched supply is designed to tolerate these excursions without losing regulation or tripping protection.

The same wattage label can hide large differences between power supplies. Platform quality, temperature rating, protection calibration, cable gauge and transient response all matter. Age and heat can reduce the headroom of an old unit.

Inspect the Physical Power Path

Shut the PC down, switch the PSU off and remove mains power. Check each plug at the GPU and supply side. It should be inserted to the connector body with no visible gap. Look for discolouration, melted plastic, damaged terminals or a cable bent hard against the plug.

Use the PSU maker's supplied modular cables for that exact model. Modular pin layouts are not universal. Where a card has several conventional PCIe sockets, check both manuals for the required number of dedicated cable runs. One daisy-chain lead will not suit every load.

Avoid pressure from a side panel on a modern compact power connector. Give the cable room to curve after the plug and support the card so movement does not work the connector loose.

Separate Power From Other Crashes

An instant black screen followed by a full restart or a system that stays off can suggest power protection. A game closing to the desktop, a driver recovery message or visual corruption may point elsewhere.

Restore GPU and CPU tuning to stock. Check Windows reliability history and event logs, while remembering that an unexpected-power event records the shutdown rather than proving its cause. Test another heavy game and a bounded benchmark. Stop if you smell hot plastic, see connector damage or hear electrical arcing.

Size the Replacement With Context

Add the processor, graphics card, drives, pumps, fans and USB-powered devices to the picture. A power-hungry CPU and GPU can peak together even when a game usually leans on one component.

Follow the graphics-card vendor's recommended PSU capacity as a starting point, then confirm that a quality model supplies the required connectors without unsafe adapters. An ATX 3.x supply designed for modern GPU excursions can be useful in a new high-power build, though certification alone does not replace model-specific reviews.

FAQ

Does every shutdown under load mean the PSU is too small?

No. A loose connector, overheating, unstable tuning, a faulty component or mains trouble can produce similar symptoms.

Can I reuse modular cables from another power supply?

Only when the manufacturer confirms compatibility between those exact PSU and cable models.

Will lowering the GPU power limit prove the cause?

It can be a useful diagnostic clue if shutdowns stop, but it does not identify whether capacity, cabling, the supply or the card is faulty.

Reseat the full GPU power path, restore stock settings and compare the exact card requirement with a quality PSU before repeating the load test.