Peak Boost vs. Sustained Clocks in Hardware Benchmarks

This article explains the critical differences between peak boost clocks and sustained clocks when benchmarking computer hardware like CPUs and GPUs. While peak boost represents the maximum speed a processor can hit during short bursts of activity, sustained clocks reflect actual performance under continuous, heavy workloads. Understanding how power limits and thermal thresholds impact these two metrics is essential for accurately evaluating hardware capabilities and real-world system performance.

What is a Peak Boost Clock?

The peak boost clock is the absolute maximum frequency a processor can achieve under ideal conditions. Hardware manufacturers prominently advertise this number on product boxes because it represents the best-case scenario for the silicon.

  • Duration: It typically lasts for only a few milliseconds to a few seconds.
  • Workloads: It is triggered during light, bursty tasks, such as opening a web browser, loading a file, or executing a single-threaded benchmark.
  • Limitations: A processor can only reach its peak boost if it has sufficient thermal headroom (it is cool enough) and power headroom (it is not drawing too much electricity). Usually, only one or two cores on a CPU will hit this speed at any given time.

What is a Sustained Clock?

The sustained clock is the frequency a processor can maintain indefinitely during a continuous, heavy workload. Once the initial boost period expires, the hardware drops down to a speed that balances performance with heat generation and power consumption.

  • Duration: It lasts for as long as the heavy workload continues, whether that is ten minutes or ten hours.
  • Workloads: It is observed during demanding, multi-threaded tasks like 3D rendering, video encoding, extended gaming sessions, and prolonged stress tests.
  • Limitations: Sustained clocks are heavily dictated by the quality of your cooling solution, the ambient temperature of your room, the motherboard's power delivery (VRMs), and the manufacturer's pre-programmed power limits.

Why the Difference Matters in Benchmarking

When benchmarking hardware, relying solely on peak boost clocks gives an incomplete and often misleading picture of performance.

Short Benchmarks vs. Long Benchmarks If a benchmark only runs for 30 seconds, a CPU or GPU might stay near its peak boost clock the entire time, resulting in an artificially high score. Because the component's heatsink has not yet been saturated with heat, the processor does not need to slow down. A 30-minute benchmark, however, forces the hardware to settle into its sustained clock speed, revealing how the system actually performs under prolonged stress.

The Impact of Thermal Throttling Two identical processors can have the exact same peak boost clock, yet perform vastly differently in sustained benchmarks. For example, a CPU inside a thin laptop and the same CPU in a well-ventilated desktop might both hit a 5.0 GHz peak boost. However, after two minutes of rendering video, the laptop CPU might drop to a sustained clock of 3.2 GHz to prevent overheating, while the desktop CPU sustains 4.7 GHz due to superior cooling.

Power Limits (PL1 and PL2) Modern processors use dynamic power states. In Intel CPUs, for example, PL2 (Power Limit 2) allows the chip to draw massive amounts of power to hit peak boost clocks for a set time limit. Once that timer expires, the chip drops to PL1 (Power Limit 1), a lower power state that dictates the sustained clock. Benchmarking only within the PL2 window hides the true sustained performance of the chip.

How to Measure Both Metrics

To accurately benchmark your hardware, you need to monitor both clock types simultaneously using telemetry software while running a stress test.

  1. Run a Monitoring Tool: Open a program like HWiNFO64 or MSI Afterburner before starting your benchmark.
  2. Identify the Peak Boost: Run a brief, lightly threaded test. Look at the "Maximum" column in your monitoring software to see the highest frequency achieved by your fastest core.
  3. Identify the Sustained Clock: Run a heavy benchmark (like Cinebench or 3DMark) on a continuous loop for at least 15 to 20 minutes. Reset your monitoring software's timer once the temperatures stabilize, and look at the "Average" or "Effective" clock speed column. This number is your true sustained clock.