A gaming PC that crashes mid-game is almost always failing in one of four ways: the graphics driver or GPU, the power supply, heat, or unstable memory. How it crashes tells you which. A black screen that recovers points to the GPU driver, an instant restart with no blue screen points to power or heat, and a stop code on a blue screen points to memory or drivers.
Resist the urge to start swapping parts. A new graphics card costs a lot and won't fix a power supply that's tripping under load. Read the crash first, test in a sensible order, and change one thing at a time.
What is the crash pattern telling you?
Different faults leave different fingerprints. Match what you see to the closest row below, then run the test in the right-hand column before spending anything.
| What happens | Likely cause | First test |
|---|---|---|
| One game closes to the desktop with an error, others are fine | Game files, mods, overlays or a corrupt shader cache | Verify game files, disable overlays |
| Screen goes black, sound stutters, Windows says the display driver stopped responding | GPU driver timeout caused by an unstable overclock, heat or weak power | Reset the GPU to stock, clean driver install |
| PC switches off or restarts instantly, no blue screen | Power supply protection, thermal shutdown or a short | Event Viewer, temperature log, PSU swap test |
| Blue screen reading WHEA_UNCORRECTABLE_ERROR | CPU, cache or memory instability, often from overclocking or a memory profile | Return CPU and RAM to default settings |
| Blue screen reading VIDEO_TDR_FAILURE, naming nvlddmkm.sys, atikmdag.sys or amdkmdag.sys | Graphics driver or GPU fault | Clean driver install, test at stock clocks |
| Everything freezes and audio loops until you hard reset | GPU or memory lock-up | Memory test, stock settings |
| Crashes only after the PC has been under load for a while | Heat building up | Log temperatures through a full session |
How do you read what Windows recorded?
Windows keeps a record of every crash, and two built-in tools show it without installing anything. Press Windows+R, type perfmon /rel and press Enter to open Reliability Monitor, which charts crashes on a timeline and names the failing app or error.
For detail, open Event Viewer and go to Windows Logs, System. Look for these entries around the time of a crash:
- Kernel-Power, Event 41: the PC restarted without shutting down cleanly. It says power was lost or the system hung, not why.
- Display, Event 4101: the display driver stopped responding and recovered. This is the signature of a GPU driver timeout.
- WHEA-Logger: a hardware error was reported by the CPU, memory or PCIe bus.
- Disk, Events 7, 11 or 51: storage trouble, not graphics.
How do you rule heat in or out?
Install a free monitor such as HWiNFO with logging switched on, then play until the game crashes. Check the log for CPU package temperature, GPU temperature, GPU hotspot, memory junction temperature and fan speeds. A single high number isn't a fault. Some processors are designed to run close to their limit, so check your chip maker's published maximum. The warning sign is a temperature that climbs steadily and ends at the crash.
If temperatures are the problem, the cause is usually mundane. Dust packs the heatsink fins and filters, a fan has stopped or slowed, thermal paste has dried out, or case airflow is blocked. Switch off, unplug, hold each fan still while you blow dust out with compressed air, and confirm every fan spins when you power on. Mumbai's monsoon humidity combined with dust can clog a case faster than you'd expect. For a professional clean, see our gaming desktop cleaning service.
Is the graphics card to blame?
Work through the cheap tests first.
- Remove every overclock and undervolt. Reset your tuning software to defaults. Many "bad GPU" crashes disappear at stock clocks.
- Reinstall the driver cleanly. Use Display Driver Uninstaller in Safe Mode, then install the latest stable driver from the chip maker's site.
- Check the power cables. Every PCIe power plug must click fully home. On high-power cards use a separate cable from the supply for each connector rather than a daisy chain, and follow the card maker's guidance on the newer 12-pin connector, which must be fully seated and not bent sharply near the plug.
- Look for artefacts. Coloured dots, checkerboard patterns or flashing textures before a crash suggest a memory or GPU hardware fault. Stop stress testing if you see them.
- Test elsewhere. Try the card in another PC, or try another card in yours.
If the card itself is failing, we confirm the fault and advise on replacement or upgrade options. Our guide to graphics card artefacts and crashes explains the symptoms in more detail, and the gaming PC GPU upgrade page covers the next step.
Could the power supply be the weak link?
Quite often. Modern graphics cards can spike far above their average power draw for fractions of a second. A supply that is old, low quality or simply too small trips its over-current protection and the PC cuts out, typically at the busiest moment of a game. Signs include shutdowns in intense scenes, a clicking or buzzing sound, a burnt smell, or a supply that was chosen for an older, less power-hungry graphics card.
The reliable test is to swap in a known-good supply of adequate wattage, not to guess from a number on a label. Check the graphics card maker's recommended supply wattage for your card. Our list shows a 650W supply at ₹3,000 to ₹5,000 and an 850W or larger unit at ₹5,000 to ₹8,000, indicative. One safety rule: never reuse modular cables from a different supply. Connectors can look identical while the wiring differs, and the wrong cable can destroy parts. Our guide on failing power supply warning signs goes deeper. Power cuts and voltage dips matter here too, so read about surge protection and UPS options in Mumbai.
Could memory, storage or Windows be involved?
Memory profiles such as XMP or EXPO run RAM faster than its default, and not every combination of sticks and motherboard is stable at that speed. Switch the profile off and test at default speed. Run Windows Memory Diagnostic for a quick check, or MemTest86 from a USB stick for a thorough one, and test sticks one at a time if you see errors. Also scan Event Viewer for disk errors and check that your system drive isn't nearly full.
In what order should you test?
- Note the symptom and how long the PC had been running when it crashed.
- Check Reliability Monitor and Event Viewer for the matching entry.
- Reset all overclocks, undervolts and memory profiles to default.
- Reinstall the graphics driver cleanly.
- Log temperatures through a full crashing session.
- Clean dust and reseat every power and PCIe cable.
- Test the memory.
- If crashes continue, swap-test the power supply, then the graphics card.
Frequently asked questions
Why does my PC crash in only one game?
A single-game crash usually points to that game's files, mods, overlays or shader cache, or to a driver bug. If every demanding game crashes, suspect heat, power or the GPU.
Can a bad power supply damage my graphics card?
It can. Protection circuits normally trip first, but a failing supply can pass unstable voltage to other parts. If you suspect the supply, stop gaming on it until it has been tested.
Is it safe to keep playing while I troubleshoot?
Short test sessions are fine. Repeated hard shutdowns can corrupt Windows files and stress the drive, so don't keep gaming through crashes.
Should I just replace parts one at a time?
That gets expensive fast. A technician with known-good spares can swap-test a supply or card in minutes, which is usually cheaper than buying parts on a hunch.
If crashes continue after the steps above, we can diagnose the machine and give a written quote before any work. Book a free pickup, or see our gaming PC repair page. The inspection fee is ₹350 and is refunded if you go ahead with the repair.
