NVIDIA Inspector Overclock Guide
Learn how to safely overclock your NVIDIA graphics card using NVIDIA Inspector. This step-by-step guide covers core clock, memory clock, voltage control, and stability testing for both RTX and GTX GPUs.
What Is GPU Overclocking?
GPU overclocking is the process of increasing your graphics card's clock speeds beyond the factory default settings. By raising the core clock (GPU frequency) and memory clock (VRAM frequency), you can extract more performance from your existing hardware — getting higher FPS in games without buying a new GPU.
NVIDIA Inspector makes overclocking easy by providing a simple interface with offset sliders. Unlike hardware modifications, software overclocking through NVIDIA Inspector is completely reversible — if something goes wrong, the driver simply resets to default clocks.
Before you start overclocking, make sure your GPU has adequate cooling. If your GPU already runs above 80°C at stock settings, consider improving your case airflow or adjusting your fan curve before overclocking.
Recommended Safe Overclock Values by GPU
How to Overclock — Step by Step
- Download NVIDIA Inspector
Get the latest version from our download page. Extract the zip file to any folder. No installation needed.
- Launch nvidiaInspector.exe
Double-click the executable. The main window shows your GPU info: name, driver version, clock speeds, temperature, and fan speed.
- Click "Show Overclocking"
At the bottom of the main window, click the "Show Overclocking" button. This reveals the overclocking panel with sliders for core clock, memory clock, and voltage.
- Set Core Clock Offset
Start conservative: set the Base Clock Offset slider to
+50 MHz. This tells the GPU to run 50 MHz faster than the factory boost clock. - Set Memory Clock Offset
Set the Memory Clock Offset slider to
+100 MHz. Memory is generally more tolerant of overclocking than the core. - Click "Apply Clocks & Voltage"
The overclock is applied immediately. You should see the clock speeds increase in the main monitoring window.
- Run a Stress Test
Open Unigine Heaven, FurMark, or 3DMark and run for at least 15 minutes. Monitor temperatures and watch for visual artifacts (flickering pixels, wrong colors, screen glitches).
- Increase Gradually
If the stress test passes with no artifacts and temps stay below 83°C, increase core by +25 MHz and memory by +50 MHz. Repeat the stress test. Keep going until you hit instability, then back off by 25 MHz.
- Test in Real Games
Run your favorite demanding game for 30+ minutes. Synthetic benchmarks and real games can behave differently. If you see any crashes or artifacts in-game, reduce your overclock.
Automate Overclocking with a .bat Script
You can create a Windows batch file to apply your overclock automatically on boot. Create a text file, paste the script below, save as overclock.bat in your NVIDIA Inspector folder, and add it to your Windows Startup folder.
@echo off
REM === NVIDIA Inspector Auto-Overclock Script ===
REM Adjust values below for your specific GPU
SET CORE=100
SET MEMORY=300
SET VOLT=0
REM Apply overclock to GPU 0
nvidiaInspector.exe -setBaseClockOffset:0,0,%CORE% -setMemoryClockOffset:0,0,%MEMORY%
echo Overclock applied: Core +%CORE% MHz, Memory +%MEMORY% MHz
timeout /t 3Win+R, type shell:startup, and move the shortcut there to auto-apply your overclock every time Windows boots.
Overclocking FAQ
Common questions about GPU overclocking with NVIDIA Inspector.
Yes, software overclocking with NVIDIA Inspector is generally safe. The tool applies offset-based overclocking which stays within the GPU's power and thermal limits. If the overclock is unstable, the driver will crash and reset to default clocks — no permanent hardware damage occurs.
Most NVIDIA GPUs can handle +100 to +150 MHz on the core clock. Start at +50 MHz and increase by 25 MHz increments while testing stability. RTX 30/40 series cards typically have less overclocking headroom due to aggressive factory boost clocks.
Memory overclocking is generally safer than core overclocking. Most GPUs can handle +200 to +500 MHz on memory. GDDR6X memory (RTX 3080, 3090, 4080, 4090) may show diminishing returns above +300 MHz due to error correction.
Run a GPU stress test like Unigine Heaven, FurMark, or 3DMark for at least 15-20 minutes. Watch for visual artifacts (flickering, colored pixels), driver crashes, or sudden temperature spikes above 85°C. If any occur, reduce your overclock.