Gamer Fixes RX 5700 XT Overheating With a Second Fan, Ends Stutters
A PC gamer reported that his graphics card started to stutter under load, with fans ramping to maximum speed and triggering multiple emergency shutdowns. After throttling set in—his clock rate was reduced down to 1,680 MHz—he decided to overhaul the card’s cooling himself. The result: notably lower VRAM temperatures and, per his report, no more stuttering in games.
What went wrong and why he intervened
He said the card’s performance quickly became unreliable, showing visible stutters while the cooling system spun at the highest possible level. The situation escalated to several emergency shutoffs. Although the card could theoretically boost as high as 2.1 GHz, his configuration ended up pulling back to a 1,680 MHz takt.
That thermal behavior matters because older graphics cards often begin reducing performance when temperatures climb. He also highlighted a practical threshold: when the hotspot temperature rises above 100 °C, it’s time to take action.
Quick facts
- Reported symptoms: game stutter, fans at maximum, multiple emergency shutdowns
- Observed behavior: clock capped to 1,680 MHz (boost could reach up to 2.1 GHz in theory)
- Cooling changes: replaced heat pads, reapplied thermal paste, added a second heatsink on VRAM using cable ties, mounted an original AMD fan to blow directly over components
- Software tweaks: reduced GPU voltage from 1.16 V to 1.07 V, raised memory clock to 1,825 MHz, increased power limit by 15%
- Fan curve: full speed targeted starting at 75 °C
- Temperature results: VRAM dropped from 104 °C to 85 °C; hotspot dropped from 106 °C to 83 °C
- Outcome claimed: stuttering stopped
Step-by-step: the cooling rebuild
His first move was maintenance-level: he replaced heat pads that had become visibly brittle and renewed the thermal paste between the GPU and the cooler. However, he didn’t see the major performance gains he was hoping for, so he went further.
He removed the backplate and then attached a second heatsink directly to the VRAM chips, securing it with cable ties. To improve airflow, he installed an original AMD fan designed to push cool air over the components while carrying warm exhaust toward the case’s I/O openings.
Undervolting and power/memory adjustments
Hardware cooling wasn’t the only lever he pulled. In MSI Afterburner, he lowered the voltage from 1.16 volts to 1.07 volts. He also increased the memory clock to 1,825 MHz and raised the power limit by 15%.
To keep temperatures from creeping back up, he adjusted a custom fan profile so that the fans should reach their highest RPM once the temperature hits 75 °C. With the combined changes—cooling modifications plus these tuning settings—he reported a major temperature reduction on both VRAM and the hotspot.
Measured temperature drop and reported gameplay improvement
After all the tweaks, his VRAM temperature fell from 104 °C to 85 °C. The hotspot temperature improved even more strongly, dropping from 106 °C to 83 °C.
He stated that games no longer stutter after the work was completed. The broader point he made is that heat-related throttling can make a card appear “done,” but hands-on cooling fixes—done with the right know-how—may extend usable life by preventing the thermal triggers that force down performance.
Why this matters for older GPUs
Thermal throttling is common as graphics hardware ages, especially when hotspot temperatures climb high enough to force protective behavior. His warning was direct: if the hotspot passes 100 °C, it’s a clear sign to investigate cooling and contact quality.
He also framed the repair as a potential alternative to replacing the GPU outright—provided the work is performed carefully, since opening and modifying a graphics card carries real risk if you’re not experienced.
