Quick Answer
Lost FPS on an RX 9070 XT build is usually thermal throttling, a stale driver, or a background hog, not a dead card. Watch GPU temp in HWiNFO64 (throttling starts near 90C), reinstall a clean AMD Adrenalin 25.x driver, and check the CPU is not pinned at 100%. Most rigs recover full frames after a clean driver pass.
Read the logs before you swap anything
Open Event Viewer (Windows Logs then System) and look for Kernel-Power 41 (a dirty restart), WHEA-Logger 18 and 19 (a hardware error), and Display TDR 4101 (a driver timeout). Sort by the timestamp of the last fault so you compare like with like. In HWiNFO64, log a five-minute gaming run and check whether the FPS drops you see coincide with a temperature spike, a clock drop, or a voltage dip. A reboot or freeze wipes the screen evidence, so the log is the only honest record of what actually happened.
The fix sequence that actually works
Run DDU in Safe Mode, then reinstall a clean AMD Adrenalin 25.x driver. In the motherboard firmware, enable Resizable BAR (Smart Access Memory) and confirm the card is in the top PCIe x16 slot. Confirm a 750W-class power supply such as a Corsair RM750e or Cooler Master MWE 750, and reseat the power connector fully. Watch temperatures: idle should sit around 35-44C, sustained load around 65-75C, and throttling kicks in near 90C.
Where SA buyers usually go wrong
Most call-outs we see locally are not faulty parts at all. Dust build-up from a dry Highveld winter, a cheap unbranded cable, a half-seated power connector, or a stale driver after a feature update cover the bulk of them. Before you budget for a replacement, work the free fixes first: a clean driver pass, a CMOS reset, a dust-out, and a known-good cable cost nothing. If a part genuinely is faulty, every component sold through Evetech carries a local warranty, so you log the RMA here rather than chasing an overseas seller. A quality power supply is the cheapest insurance: a 750W unit from around R1,500 protects a build worth far more.
FAQ
Why did my FPS drop without changing anything?
The usual causes are thermal throttling (watch for the part passing 90C in HWiNFO64), a stale driver after an update, or a background process pinning the CPU. A clean driver pass and a temperature check recover most lost frames.
How do I know if it is the CPU or GPU?
Log both in HWiNFO64 during a session. If GPU usage drops below 95% while FPS falls, the CPU or a background task is the limit; if the GPU stays pinned and clocks fall, it is heat or power.
Will a clean driver re-install really help?
Often, yes. Removing the old driver with DDU and installing a clean AMD Adrenalin 25.x driver clears corrupted shader caches and leftover profiles that quietly cost frames after a feature update.
Fix the cause, not the symptom.
Before you shop for parts, run DDU and reinstall a clean AMD Adrenalin 25.x driver on a clean session, then re-test on one repeatable scene. If a part is genuinely faulty, log the local warranty through Evetech rather than guessing.