How G-Sync and FreeSync Impact Hardware Benchmarking

Variable refresh rate (VRR) technologies like Nvidia G-Sync and AMD FreeSync are designed to create a smooth visual experience, but they actively interfere with accurate computer hardware benchmarking. This article explains how display synchronization alters frame delivery, enforces artificial performance ceilings, and skews frame time metrics, demonstrating why these features must be disabled when testing the raw performance of a CPU or GPU.

The Conflict Between Smoothness and Raw Performance

The primary goal of hardware benchmarking is to push components to their absolute limits to measure their maximum possible output. Conversely, the goal of G-Sync and FreeSync is to control and stabilize frame delivery to eliminate screen tearing and stuttering.

VRR achieves this by dynamically matching the monitor’s refresh rate to the graphics card's frame rate output. Because benchmarking requires unconstrained hardware behavior to compare different components fairly, introducing a technology that actively regulates frame output invalidates the test results.

Artificial Frame Rate Caps

The most significant impact VRR has on benchmarking is the introduction of frame rate limits. G-Sync and FreeSync only function within a specific refresh rate window, typically up to the monitor's maximum refresh rate (e.g., 144Hz or 240Hz).

To keep the frame rate within this functional window, VRR technologies are almost always paired with V-Sync or automatic driver-level frame limiters (such as Nvidia Reflex). If you are benchmarking a high-end GPU capable of rendering 350 frames per second (FPS) in a specific game, an active VRR setup on a 144Hz monitor will cap the performance at or just below 144 FPS. Consequently, the benchmark will only measure the monitor's limitation rather than the GPU's actual power.

Skewed Frametimes and 1% Lows

Modern benchmarking relies heavily on frametime metrics—specifically 1% and 0.1% lows—to measure system stability and stutter. Frametime is the exact amount of time it takes the hardware to render a single frame.

When G-Sync or FreeSync is enabled, the graphics driver communicates with the display to manage the render queue. If the frame rate hits the monitor's maximum refresh rate ceiling, the system stops rendering new frames and waits for the display to catch up. This introduces artificial delays in the rendering pipeline, which benchmarking software records as higher frametimes. This makes the hardware appear slower or less consistent in data logs than it actually is when running uncapped.

Altered CPU and GPU Utilization

Benchmarking often involves isolating either the CPU or the GPU to see where a system bottleneck lies. Because VRR setups frequently prevent the hardware from running at 100% capacity when the frame rate approaches the monitor's refresh limit, system utilization drops.

This results in lower power consumption, reduced thermal output, and lower clock speeds during the test. If you are benchmarking a cooling solution, power supply, or overclocking stability, VRR will prevent the hardware from generating the maximum heat and power load required for an accurate stress test.

Inconsistent Cross-System Comparisons

A core principle of benchmarking is repeatability. If you benchmark a graphics card on a 144Hz FreeSync monitor and compare the results to the exact same graphics card tested on a 240Hz G-Sync monitor, the results will differ wildly if synchronization is left on. The hardware did not change, but the display constraints dictated the benchmark scores.

How to Properly Configure a System for Benchmarking

To measure the true performance of your computer hardware, you must remove the display as a limiting factor. Before running any benchmark, apply the following settings:

  • Disable G-Sync or FreeSync in the Nvidia Control Panel, Nvidia App, or AMD Radeon Software.
  • Disable V-Sync both in the graphics driver control panel and within the specific game or benchmark settings.
  • Remove all frame rate limiters in the game menu and third-party software (such as RivaTuner Statistics Server).
  • Disable Low Latency Modes like Nvidia Reflex or AMD Anti-Lag, as these features can dynamically cap frame rates when paired with VRR displays.