CPUs slow down during heavy use to save energy and prevent overheating.
CPUs slow down during heavy use to save energy and prevent overheating.
Initially, if a processor is labeled for a peak boost frequency of X GHz, I've never observed it hitting that speed. For instance, my older computer's chip was built to handle a maximum turbo of 2.70 GHz, yet I've never noticed it reaching such a rate. The best I've recorded is around 2.67 GHz, and even then it happened only under certain demanding tasks that required the highest clock speeds or intense single-threaded performance without excessive strain so it wouldn't throttle. Why do CPUs slow down under load? Not because of heat or power constraints. In fact, I've never seen a CPU automatically reduce speed just because of workload—it's usually due to power or thermal restrictions.
Typically, this happens because the boost clock applies to one core. When a task uses multiple cores, it doesn't hit that boost. Additionally, power or temperature issues often cause CPUs to downclock.
The speed boost circuits are designed for single-core processors, not all-in-one setups. A CPU's energy and heat management capabilities are constrained, meaning when maximum load is applied with all cores requesting power, the system must share this workload across every core rather than concentrating it on one. This affects both thermal performance and power usage. Enhancing cooling solutions and increasing power supply can help achieve higher all-core speeds. Additionally, laptops have built-in limits on how long they can sustain boost mode to stay within safe temperature ranges.
Without increasing the frequency, not every core will display the highest possible speed. Intel once openly shared detailed charts called "Boost Tables" showing expected speeds for different numbers of cores under stress. For instance, the first generation i7 870 boosts to its rated 3.6GHz only when one core is active; with three or four cores, it caps at 3.2GHz. These tables are still available today through software such as Intel XTU, though Intel no longer makes them easily accessible. AMD’s approach is less transparent—there’s no standard Boost Table, and their method relies on factors like temperature, power usage, and voltage. Nonetheless, in real-world use, you’ll typically see the full rated boost clock when all cores are running.
There are several factors involved. First, power usage efficiency matters. Pushing a CPU beyond its optimal speed range consumes significantly more energy, so restricting maximum clock speeds helps keep power consumption within safe limits. Second, thermal management becomes critical. Ignoring efficiency and focusing solely on performance makes cooling much more challenging. Standard solutions like box coolers would need to be substantially upgraded—similar to how AMD added a robust AIO when launching the FX-9590. Third, chip design constraints play a role. The highest boost speed isn’t chosen randomly; software identifies which cores perform best. Not all cores can reach full speeds at default voltages, and requiring every core to do so could cause reliability issues. By capping all-core boost, more components meet performance standards and can be marketed as having a higher core count.
Typically clocks are adjusted when time zones shift for a few reasons. Protection from heat, avoiding damage to power sources, or limiting exposure time are common causes. If you haven’t experienced a full boost, controlling it might be difficult. This phenomenon is known as thermal runaway, which can occur under certain cooling conditions.