Is your GPU running hotter than it should, louder than you can tolerate, or drawing more power than your build was designed to handle? Knowing how to underclock your GPU is the practical answer — and it takes less than 20 minutes to do correctly. This guide covers the full process: how to decide whether underclocking is right for your situation, a step-by-step method for beginners, an advanced approach for users chasing maximum efficiency, and how to maintain a stable setup long-term. For more hardware guides, check out the laptops section.
Underclocking sets your GPU's operating frequency below the manufacturer's default. The GPU still works — it just works at a more controlled, efficient pace. You get lower temperatures, reduced power draw, and quieter fans without buying new hardware. Unlike overclocking, which pushes components past their rated spec and introduces real instability risk, underclocking works with the GPU's design. Every change is reversible with a single click.
The standard tool for this is MSI Afterburner — free, works on both NVIDIA and AMD cards, and widely supported. You adjust a clock slider, save a profile, run a stability test, and measure the results. That's the whole process. The rest is knowing when it's worth doing and what to watch for afterward.
Contents
Underclocking is a targeted fix, not a universal upgrade. Before adjusting anything, confirm that it actually solves your problem.
These are the scenarios where underclocking delivers clear, measurable improvements:
Don't underclock if your GPU temperatures are already healthy (under 80°C under load) and you're chasing the highest possible frame rates in demanding games. Every MHz you remove is compute you're leaving on the table, and if thermals aren't the problem, there's no benefit to justify the trade.
Don't underclock as a substitute for diagnosing an actual problem. A GPU that runs hot because of dust buildup or dried thermal paste needs cleaning and repasting — not a reduced clock. Fix the root cause first. And don't underclock a GPU-bottlenecked workload like heavy 3D rendering; you'll just extend your render times without any compensating benefit.
There's a wide spectrum between "adjust one slider and call it done" and "hand-tune a voltage-frequency curve for maximum efficiency." Here's how both approaches work and which one fits your situation.
This is the correct starting point for the vast majority of users. You're adjusting one variable, the feedback is immediate, and the risk is essentially zero.
Step 1 — Download and install MSI Afterburner. Get it from MSI's official site. Accept the bundled RivaTuner Statistics Server (RTSS) installation — it provides a real-time GPU stats overlay while you're gaming or stress-testing. Once installed, open Afterburner and take a screenshot of the default values before touching anything. That's your recovery baseline.
Step 2 — Establish your baseline. Run a demanding workload for 10–15 minutes — a GPU benchmark like FurMark or 3DMark, or a game you play regularly. Note your peak GPU temperature, core clock speed under full load (shown in MHz in Afterburner's monitoring tab), and fan speed percentage. You need this data to measure whether your underclock actually worked.
Step 3 — Lower the core clock speed. In Afterburner, find the Core Clock (MHz) slider and move it down by 50–100 MHz from the default. Click the checkmark to apply. No reboot needed — the new speed is immediate. If you also want to reduce memory clock, use the Memory Clock (MHz) slider with the same conservative reduction. Save your adjusted profile to one of Afterburner's numbered slots.
Pro tip: Start with a 50 MHz reduction, not 200. Small increments let you identify your GPU's stable floor without accidentally pushing past it on the first try.
Step 4 — Test stability and compare results. Run the same benchmark you used in Step 2 for at least 30 minutes. Watch for crashes, visual artifacts (flickering textures, black screens), or driver errors. If everything runs cleanly, compare your new peak temperature to the baseline. Then decide whether you want to push the underclock further. Reduce in 25–50 MHz increments each time and retest before going lower. Finding the stable floor is iterative — take your time.
Advanced users edit the GPU's voltage-frequency curve directly instead of applying a flat clock offset. In MSI Afterburner, press Ctrl+F to open the curve editor. This lets you define the exact voltage at which your GPU operates at a given frequency — for example, locking in 1750 MHz at 900mV instead of whatever voltage the factory default uses at that speed.
The result is a GPU that runs significantly cooler and more efficiently across the entire load range, not just at peak. This technique, a controlled form of dynamic voltage scaling, is what enthusiasts use when they want maximum thermal efficiency without sacrificing performance. Temperature reductions of 15–20°C are achievable on most mid-range and high-end GPUs without any noticeable frame rate loss. The trade-off is time — proper curve tuning takes an hour or more of testing across different workloads.
If you're new to GPU tweaking, do the 4-step beginner method first. The advanced approach builds on that foundation — you need to know how your card responds to clock changes before you start manipulating voltage curves.
Theory is useful. Specific numbers are better. Here's what GPU underclocking actually produces in practice.
For most gaming scenarios, a conservative underclock of 100–150 MHz produces a frame rate reduction of 2–5%. On a GPU running at 120 fps, that's the difference between 120 and 115 fps — imperceptible in actual gameplay. Meanwhile, sustained GPU temperature under full load typically drops 8–12°C, fan noise decreases to a lower RPM tier, and peak power draw falls noticeably.
| Underclock Level | Core Clock Reduction | Avg Temp Drop | Approx FPS Impact | Fan Noise Change |
|---|---|---|---|---|
| Mild | 50–100 MHz | 5–8°C | <2% | Slight reduction |
| Moderate | 100–200 MHz | 8–14°C | 2–5% | Noticeable reduction |
| Aggressive | 200–300 MHz | 12–20°C | 5–12% | Significant reduction |
| Voltage Curve (Advanced) | Variable | 15–22°C | 0–3% | Major reduction |
Pairing the right GPU with the right processor also determines how much headroom you have for underclocking without bottlenecking the system. Our guide to the best CPUs for the 1080 Ti covers the processor-GPU combinations that give you the most flexibility.
Home theater PCs are ideal candidates for aggressive underclocking. Streaming 4K video and playing local media files demand a small fraction of a modern GPU's compute capacity. Running your GPU at 60–70% of its stock speed during these workloads reduces power draw substantially, keeps the system near-silent, and extends the component's operational life. Our roundup of the best video cards for HTPC builds includes models that respond particularly well to this approach — cards designed for low thermal envelopes that perform even better when underclocked.
For creative workstation tasks like video rendering or 3D modeling, the equation shifts. These workloads saturate GPU compute resources, so aggressive underclocking will extend render times proportionally. A mild 50–100 MHz reduction still cuts thermals without meaningfully impacting your queue — but this isn't where you push the underclock to its limit.
A well-configured underclock reduces thermal stress, but it doesn't replace routine hardware care. A dust-clogged cooler running an underclocked GPU still runs hotter than a clean cooler running at stock speeds.
Keep MSI Afterburner's hardware monitoring overlay active. After setting your underclock, regularly check these metrics under full load:
Maintenance note: Reapplying thermal paste to an aging GPU every 2–3 years can reduce temperatures by 10–15°C on its own — sometimes more than underclocking achieves without touching clock speeds at all.
Physical cleanliness matters as much as software settings. Compressed air through your GPU heatsink every few months keeps the cooling solution performing at rated capacity. Dust accumulation is cumulative and silent — your GPU won't announce that it's losing 8°C of cooling headroom from buildup.
GPU behavior changes over time. Thermal paste degrades. Fan bearings wear. Ambient temperatures shift seasonally. Check your underclocked profile every few months and compare current temperatures against your original baseline. If temps are creeping upward despite the same clock settings, your cooling hardware needs attention — not a deeper underclock.
Always retest your underclocked profile after major GPU driver updates. Driver changes occasionally affect how the card responds to manual clock offsets, and a profile that was perfectly stable before may exhibit instability after a significant driver revision. A 30-minute stress test after each major driver update is the right habit to build.
Most underclocking problems fall into a small number of categories. Each one has a clear cause and a straightforward fix.
A black screen, driver crash, or full system reboot during a stress test means one thing: your underclock is too aggressive for the current workload. You've pushed the core clock below the GPU's stable operating floor at that load level. This is not a hardware problem — it's expected behavior when you're finding the limit through iteration.
The fix is simple: reboot, open Afterburner, and raise the core clock by 25–50 MHz. Retest for 30 minutes. If stable, you've found a workable floor. If it crashes again, raise the clock another 25 MHz. Repeat until the system runs cleanly. No permanent damage occurs from an unstable underclock — the GPU resets when the driver crashes or the system reboots.
If your frame rates feel worse after underclocking, two things are likely happening. First, you may have set the underclock too deep — revert to a milder setting and compare frame rate benchmarks objectively rather than by feel. Second, your GPU may now be CPU-bottlenecked; if you reduced GPU performance significantly in a CPU-limited game, the GPU spends more time waiting and the overall experience degrades.
Run a proper benchmark (3DMark, Unigine Heaven) before and after your underclock changes to get objective data. Feel is unreliable. Numbers tell you exactly what's happening and by how much.
Misinformation keeps many users from a change that would genuinely improve their systems. Here are the two most persistent myths about GPU underclocking — and the facts that debunk them.
This is the most frequently repeated piece of incorrect GPU advice online. Underclocking does not void your warranty. Overclocking beyond factory spec is what voids warranties at many manufacturers — because it pushes components outside their tested operating range. Underclocking does the opposite: it reduces operating speed, lowers electrical stress, and decreases wear. GPU manufacturers designed their products to run within a range of clock speeds, and operating below the advertised boost clock is explicitly supported behavior. The fact that MSI — a GPU manufacturer — provides Afterburner as an official tool confirms this.
This argument assumes the performance trade-off is large. For a typical moderate underclock of 100–150 MHz, the actual gaming frame rate impact is 2–4%. On a 120 fps workload, that's 2–5 frames per second. In exchange, you get a 10–14°C temperature reduction, noticeably quieter operation, and lower power draw. Most users who run this experiment decide the trade is absolutely worth it — especially in thermally-constrained builds where the alternative is thermal throttling that already costs frame rates dynamically.
Advanced voltage curve tuning eliminates even this modest trade-off. You recover the efficiency lost to factory-conservative voltage settings, and the GPU runs cooler with no frame rate reduction at all. The performance argument against underclocking assumes you never tune beyond the simplest method — and that assumption doesn't hold.
No. Underclocking reduces operating speed, which lowers heat output and electrical stress on the GPU's components. It's the thermal opposite of overclocking, which does carry damage risk when pushed beyond stable limits. A correctly tested underclock reduces component wear over time rather than increasing it — lower sustained temperatures extend the working life of the silicon and VRM components.
Start with a 50–100 MHz reduction below the stock core clock and test for 30 minutes of heavy load. Most GPUs tolerate 100–200 MHz reductions without any instability. Beyond 200 MHz, results vary by card and workload. Use the iterative method — reduce by 25–50 MHz increments, test each step fully — rather than jumping straight to a large offset. Your GPU's stable floor is card-specific and workload-dependent; testing is the only reliable way to find it.
Yes, in most cases. Thermal throttling happens when your GPU exceeds its programmed temperature limit and automatically reduces its own clock speed to cool down. By setting a lower manual clock ceiling before that threshold is reached, you give the GPU stable headroom to operate within. The result is consistent frame rates instead of the stuttering performance that comes from the GPU bouncing between its boost clock and its throttled speed under sustained load.
The best GPU setting isn't the highest one your card can hit — it's the lowest one it can sustain quietly, stably, and indefinitely.
About Malcolm Woods
Malcolm Woods is a technology writer and sustainability advocate with a background in consumer electronics and a long-standing interest in the intersection of technology and environmental impact. He has spent years evaluating tech products — from smartphones and smart home devices to solar-powered accessories — with a focus on real-world performance, longevity, and value. At the site, he covers tech accessory reviews, smart home gear, buying guides, and practical how-to content for everyday technology users.
Go for the FREE Gifts. Or check out for free energy books from our best collection.
Remove Ad block to reveal all the secrets. Once done, hit a button below