Diagnosing and Fixing CPU Thermal Throttling

What Thermal Throttling Actually Is

Modern CPUs monitor their own temperature continuously and reduce clock speed automatically when temperatures approach a hard limit (usually 95-100°C). This is protection, not failure — the CPU is preventing damage to itself. But it also means the CPU is running below its potential performance, which shows up as lower framerates in games and longer render times in productivity workloads.

Throttling can be constant (temperatures always at the limit during load) or intermittent (brief throttles during load spikes). Both are worth fixing. Constant throttling usually indicates a cooling problem that needs hardware attention. Intermittent throttling can sometimes be resolved with fan curve tuning or a modest undervolt.

Detecting Throttling

HWiNFO64 (free, hwinfo.com) is the reference monitoring tool on Windows. Install it, open the Sensors view, and log during a load test. The critical values to watch: CPU package temperature, effective clock speed per core, and any "throttling" flag the tool exposes (Intel calls it "PROCHOT active"; AMD calls it "TDC/EDC/TCTL limits").

Run Cinebench R23 or Prime95 small-FFT for 10-15 minutes while logging. Compare the effective clock speed under sustained load to the CPU's rated base and boost clocks. If a Ryzen 7 with a 4.5 GHz base is running at 4.0 GHz during Cinebench, it is throttling. If a 5.5 GHz-boost CPU never exceeds 5.0 GHz under any load, it is likely thermally limited.

Games are less reliable for throttling detection because framerate is affected by many things beyond CPU. Use synthetic benchmarks or long compilation runs for reliable throttling identification, then verify game performance improves after fixing.

Root Causes in Order of Frequency

Most common: undersized or poorly-installed cooler. A stock cooler on a mid-tier CPU frequently throttles under sustained load; the fix is an aftermarket cooler. A properly-installed aftermarket cooler that still throttles usually has a mounting problem (uneven pressure, wrong paste application) rather than the wrong cooler choice.

Second most common: dust-choked cooler. A CPU cooler that was working fine 18 months ago can throttle now if the fins are packed with dust that reduces airflow through the cooling surface. Compressed air cleaning restores original performance.

Third: bad case airflow. The CPU cooler cannot cool below the temperature of the air it receives. A case with hot ambient air (poor airflow, top-intake fans, dust-blocked filters) starves even a good cooler of the cold air it needs.

Fourth: aged thermal paste. Paste over 4-5 years old on a heavily-used CPU can dry out or pump out from the mounting area, creating voids in the thermal joint. Repasting restores contact quality.

Fifth: high ambient room temperature. A 35°C summer room raises every internal temperature by roughly the same delta. If the PC only throttles in summer, this is the cause — solutions include better room cooling, better case airflow, or an undervolt to reduce the CPU's heat output.

Sequenced Fixes

Work through fixes cheapest first. Step 1: clean the cooler and case with compressed air. Step 2: verify the cooler mount is correct (re-torque screws in cross pattern). Step 3: tune fan curves to be more aggressive under load. Step 4: verify case airflow is not obstructed. Step 5: try an undervolt through Ryzen Master, Intel XTU, or the BIOS voltage/offset settings — often reduces temperature by 5-10°C at negligible performance cost.

If those steps do not resolve it: step 6, repaste the CPU with fresh quality paste. Step 7, replace the cooler with a larger one appropriate for the CPU's sustained wattage. Step 8, if the case has poor airflow, replace the case. This last step is rarely necessary but is the right answer for anyone in a solid-front case trying to cool a flagship CPU.

Quick Buying Guide

Quick Notes on Undervolting

Undervolting means reducing the voltage the CPU requests at each clock speed. Because heat scales with voltage squared, small voltage drops produce meaningful temperature reductions. Modern CPUs vary in overhead — some can drop 100-150 mV without stability issues; others cannot. Test with a stress test after each adjustment.

Undervolting Step by Step

Undervolting is the fastest way to reduce heat output without hardware changes. For AMD Ryzen CPUs: install Ryzen Master, use the PBO Curve Optimizer with a starting offset of -15 (negative offsets reduce voltage). Test with Cinebench for stability; increase the offset (more negative) in increments of 5 until instability, then back off by 5-10.

For Intel CPUs: use Intel XTU or BIOS voltage/offset settings. Start with a -50 mV offset, test stability, increase to -75, -100 mV until instability. Modern Intel CPUs typically tolerate 75-125 mV offsets without stability issues.

The temperature reduction from undervolting is typically 5-15°C at the same performance. Combined with better cooling, undervolting can turn a throttling situation into a stable, quiet system without spending on a new cooler.

Verifying Each Fix

Test after each individual fix rather than stacking changes. Cleaned the cooler? Run Cinebench and record the temperatures. Repasted? Run Cinebench again and compare. Undervolted? Same. This lets you attribute the improvement to the correct cause and know what to expect if you have to redo any step later.

Fixes This Guide Recommends

Replacement thermal paste and a real CPU cooler — the two most common fixes for throttling problems.