Laptop Thermals

Every computer on the planet has to deal with keeping itself cool somehow. The various methods, approaches, and implementations of cooling solutions directly impact the performance potential of a computer. When it comes to desktop computers, the situation is often "the quietest cooler" and less-so "the best cooler."

When it comes to laptops, the efficiency and performance of the cooler determines things outside of just performance, such as the temperature of the outer chassis. As a result, it's important that laptops keep their chips at manageable temperatures. Plus, the heatsink needs to be efficient enough to prevent the heat from bleeding into the laptop's outer shell.

Determining Thermal Performance
First and foremost, a stress test can be used to determine how well the thermal solution in a laptop is working. Often, a 15-minute long test is enough to check for heat-soak and any odd throttling behavior. However, some weird throttling behavior can be observed in some laptops with a 30-minute stress test, such as the AMD variants of the HP OMEN 15 (2020 and Lenovo Legion 5. Regardless, a 15 minute combined CPU and GPU load should be enough to exhibit any bad throttling behavior.

Recommended tools for CPU stress testing would be Prime95 or AIDA64's stress test in CPU+FPU mode. For the GPU, Furmark or a near-constant, looped benchmark load (such as UNIGINE Heaven Benchmark at max settings) will work. The goal is to put the CPU and GPU into the highest power state possible to see how their thermal management solutions respond. HWInfo64 is recommended for monitoring sensor data, but AIDA64 also has a handy graph printout in the CPU stress test window.

Data that should be monitored during a stress test include...
 * Average CPU and GPU temperatures
 * Average power consumption on the CPU and GPU
 * Any throttling/overheating indications past the first minute of the test

For the CPU, you're looking for sustained temperatures around or below 90C while maintaining power consumption levels at or around the rated TDP of the CPU. For the GPU, the same applies for power consumption. However, for temperatures, staying below 80C is your best bet. Also, check both the CPU and GPU sensor data for any throttling switches that were switched to "YES" at any point during the test. Having this occur during the beginning of the test can be okay, but if it keeps happening during the test, this can lead to stutters and dropped frames.