Variable refresh rate (VRR) is a display capability that allows the interval between successive screen refreshes to change during operation, rather than remain fixed. In interactive computer graphics, it commonly synchronizes a display with frames produced by a graphics processing unit (GPU). By beginning a refresh when a complete frame is available, within the display’s supported timing range, VRR can reduce screen tearing and presentation irregularities without imposing all the waiting associated with conventional fixed-refresh synchronization. Related variable-refresh techniques also allow devices to lower their refresh frequency when displaying static or slowly changing content.
Refresh rate and frame rate
A display’s refresh rate is the frequency at which it refreshes its image, expressed in hertz (Hz). A fixed 60 Hz display begins a refresh approximately every 16.67 milliseconds; a fixed 144 Hz display does so approximately every 6.94 milliseconds.
Frame rate describes how frequently a source produces new images, commonly measured in frames per second (fps). Rendering time can vary as scene complexity, processor workload, or other conditions change. A game averaging 90 fps therefore does not necessarily produce one frame every 11.11 milliseconds.
With a fixed-refresh display, frame completion and refresh timing may not coincide. Two common consequences are:
- Screen tearing: if the image source changes during scanout, one displayed refresh can contain portions of different rendered frames. Moving objects or scene edges may appear discontinuous across a horizontal boundary.
- Uneven frame presentation: if complete frames are shown only at fixed refresh boundaries, a frame that misses a boundary may remain undisplayed until a later one, while the preceding frame stays visible longer.
Traditional vertical synchronization—usually called V-sync—coordinates image presentation with refresh boundaries to prevent tearing. Depending on buffering and scheduling, it can introduce waiting, repeated frames, or additional latency. VRR instead allows the next refresh boundary to move within an allowed range.
How it works
Variable refresh intervals
In common VRR implementations, the display still scans the active image in an orderly sequence. The principal adjustment is to the time between successive active scans, often by extending the vertical blanking interval. This is the interval after one active image scan and before the next begins.
The source can wait for a completed frame and then initiate another refresh, subject to the display’s timing limits. VRR does not generally mean that every pixel is updated independently or that a partially rendered frame is progressively exposed as rendering proceeds.
If successive refreshes begin at times and , the corresponding refresh interval is
and its equivalent refresh frequency is
Unlike fixed-refresh operation, can vary from one refresh to another.
For example, a frame ready after 10 milliseconds can be presented on a roughly 100 Hz cadence, while a following frame requiring 12.5 milliseconds can be presented on a roughly 80 Hz cadence. Actual presentation also depends on scheduling, transmission, and display processing.
The supported range
A VRR display operates over a specified range, such as 48–144 Hz. This corresponds to refresh intervals of approximately 20.83–6.94 milliseconds.
The supported range is not necessarily the same for every connection or display mode. Resolution, color depth, interface bandwidth, display electronics, and firmware can affect available combinations.
When rendering exceeds the maximum VRR frequency, the display cannot simply refresh faster. Behavior then depends on presentation settings: the system may limit frame production, wait for a refresh opportunity, discard queued frames, or permit tearing.
When rendering falls below the minimum VRR frequency, the display cannot indefinitely extend its refresh interval. It must refresh again even if no new frame is available.
Low frame rate compensation
Low frame rate compensation (LFC) addresses operation below the native VRR range by transmitting the same rendered frame more than once.
For example, if a display supports 48–144 Hz and a game produces 40 fps, each rendered frame can be transmitted twice, producing an 80 Hz refresh cadence. At 30 fps, repeating each frame three times can produce a 90 Hz cadence.
This does not generate additional motion information: 40 unique frames per second remain 40 unique frames per second. Repetition instead keeps the display’s refresh cadence within its supported range while preserving the timing of new rendered frames.
Effective compensation depends on the width of the refresh range and the implementation’s scheduling rules. Changes in repetition multiplier can also cause visible brightness fluctuations on some displays.
Standards and implementations
VRR is a general capability rather than a single protocol or product. Its major implementations overlap, but support for one does not automatically establish support for every other one.
| Technology | Role |
|---|---|
| **[[vesa-adaptive-sync | VESA Adaptive-Sync]]** |
| **[[hdmi | HDMI]] VRR** |
| **[[nvidia-g-sync | NVIDIA G-SYNC]]** |
| **[[amd-freesync | AMD FreeSync]]** |
VESA also operates the Adaptive-Sync Display compliance program. Such certification programs assess defined behavior and performance; they are distinct from merely supporting the underlying signaling protocol.
A connector’s advertised version alone does not establish VRR support, its operating range, or compatibility with a particular source. Individual features, supported modes, and certification requirements matter.
System integration
Successful VRR operation depends on more than the panel. The graphics hardware, display connection, device driver, display electronics, and presentation software must work together.
The operating system and its desktop compositor can influence whether frames reach the display with the required timing. Support may differ between exclusive full-screen applications, borderless windows, and ordinary desktop windows. Multi-monitor configurations can introduce additional scheduling constraints, particularly when displays use different refresh rates.
An application does not necessarily need to implement the display protocol itself. It ordinarily submits rendered frames through a graphics or presentation application programming interface, while lower-level components manage display timing. Nevertheless, application buffering, frame queues, and frame-rate limits affect the result.
VRR also interacts with latency management. Presenting a newly completed frame promptly can reduce waiting relative to some fixed-refresh V-sync configurations, but a long queue of already-rendered frames can still create substantial delay. VRR alone does not guarantee low end-to-end latency.
Applications
Games and interactive graphics
Games are a principal application because their rendering workload often varies from frame to frame. VRR allows these variations to appear without the tearing of unsynchronized presentation or the fixed-boundary repetition associated with conventional V-sync.
It does not improve the simulation itself or make slow rendering fast. A game running at 25 fps still supplies only 25 distinct images each second, even if frame repetition keeps the display operating at a much higher refresh frequency.
Video playback
VRR can allow compatible video systems to present frames at a cadence appropriate to the source. However, its usefulness depends on the display’s range and playback software.
VRR is not the only solution to video cadence mismatch. A fixed 120 Hz display, for example, can present 24 fps material with each source frame repeated exactly five times. Conversely, a video source below a VRR display’s minimum frequency may require frame repetition or another supported mode.
Mobile and power-sensitive devices
Laptops, tablets, and smartphones can adjust refresh frequency according to content and interaction. Lower frequencies may reduce display-system activity during static content, while higher frequencies support smoother scrolling and animation.
This usage overlaps with gaming VRR but is not identical. A device that automatically switches among several fixed refresh modes is not necessarily synchronizing every refresh to an asynchronously completed rendered frame. Actual energy savings depend on the panel, display controller, graphics workload, and system design.
Limitations and visual artifacts
VRR does not correct poor frame pacing. If an application produces frames at irregular intervals, the display can reproduce those intervals more faithfully, but the underlying irregular motion remains. Long stalls remain visible.
VRR does not eliminate motion blur. Blur can arise from pixel response time and from the persistence of an image while the viewer’s eyes track motion. More distinct frames, faster transitions, or specialized illumination techniques address different parts of this problem.
Pixel response behavior can vary with refresh frequency. Liquid-crystal displays commonly use overdrive to accelerate pixel transitions. An overdrive setting suitable for high refresh frequencies may produce overshoot or other artifacts at lower frequencies. Some displays adjust overdrive dynamically.
Brightness can fluctuate. Some panels exhibit luminance changes as refresh intervals vary, particularly during unstable frame delivery or transitions into and out of frame repetition. Such VRR flicker is distinct from tearing.
Some display features may interact with VRR. Backlight strobing, local dimming, and other processing modes can have restrictions or different behavior during variable-refresh operation. These are implementation-dependent rather than universal limitations.
Scanout still takes time. VRR controls when a refresh begins; it does not necessarily update the entire screen simultaneously. Pixels near the end of the scan may show a new frame later than pixels near the beginning.
Distinction from related techniques
VRR is frequently discussed alongside other performance technologies, but their functions differ:
- A high fixed refresh rate provides frequent presentation opportunities without necessarily matching their timing to completed frames.
- Frame-rate limiting controls how quickly an application produces frames rather than changing the display’s refresh timing.
- Frame interpolation or generation creates additional images; VRR schedules the presentation of available images without inherently creating new ones.
- Dynamic resolution scaling changes rendering resolution to manage workload rather than changing refresh timing.
- V-sync normally governs presentation relative to refresh boundaries; VRR makes those boundaries variable within a supported range.
These techniques can be combined. Their combined behavior depends on frame scheduling, buffering, and what happens when rendering falls outside the display’s variable-refresh range.