VRR or Variable Refresh Rate is a technology that makes the display change its refresh rate in real time, adapting to the speed of frame output by the graphics card. Instead of fixed 60 or 144 Hz, the monitor dynamically refreshes exactly when a new frame is ready, eliminating screen tearing and reducing latency.
The technology is in demand in gaming monitors, TVs with HDMI 2.1 support, laptops and even some smartphones. It is the basis of hardware standards like NVIDIA G-Sync, AMD FreeSync and the universal Adaptive-Sync. In the console segment, VRR is supported by PlayStation 5 and Xbox Series X, while in mobile devices Apple ProMotion and similar Android solutions combine interface smoothness with power saving.
The main technical problem is flickering on OLED and VA panels during sudden frequency jumps, especially on loading screens. The second nuance is incompatibility with some motion processing technologies like Black Frame Insertion when enabled simultaneously. Also, early implementations had ghosting due to aggressive panel overdrive at low frequencies. Finally, some VRR displays have a narrow operating range, for example 48 to 60 Hz, which when the frame rate drops below the threshold triggers Low Framerate Compensation, increasing latency.
How VRR works
The principle of VRR operation is based on changing the time interval of the vertical blanking interval while keeping the horizontal scan constant. The graphics card, having received data from the display about Adaptive-Sync support via the DisplayPort or HDMI protocol, stops sending frames on a strict timer. Instead, it holds the finished frame in the buffer and outputs it to the screen only after receiving a signal from the monitor that it is ready to refresh. This is radically different from classic V-Sync, where the graphics card artificially limits its framerate to a value that is a multiple of the display scan rate, causing noticeable input lag during performance drops.
Compared to triple buffering technology, which only smooths out the unevenness of frame delivery but maintains a fixed frequency, VRR dynamically changes the physical refresh cycle of the pixels. A hardware add-on like the G-Sync Ultimate module uses a proprietary programmable chip in the monitor that manages the crystal response acceleration depending on the current refresh cycle, preventing smearing at speeds from 1 to 240 Hz. It is the full synchronization of GPU and TCON timings, not just software buffer adaptation, that allows achieving the effect of clean motion without stutters and tears.
VRR functionality
- Frame synchronization through adaptive refresh rate. VRR technology allows the display to dynamically change the vertical scan rate in real time. Instead of a fixed refresh interval, the matrix waits for a new frame from the GPU, eliminating the need for rigid binding to standard values like 60 or 144 Hz.
- Elimination of Screen Tearing. Tearing occurs when a new buffer is sent during the display scanning process. VRR suspends the panel update until the rendering of the next whole frame is complete. As a result, the monitor outputs only complete, consistent images, eliminating horizontal displacement artifacts without activating classic vertical sync.
- Minimizing Input Lag. Traditional V-Sync forces the GPU pipeline to wait for a fixed display refresh window, creating a buffer queue. The adaptive frequency mechanism breaks this link. As soon as the GPU finishes building a frame, it is immediately transmitted to the panel, reducing the phase shift between user action and pixel reaction.
- Operation Outside the Fixed Range. Every VRR panel has minimum and maximum synchronization boundaries. When the frame rate drops below the threshold (for example, 48 Hz), the driver activates Low Framerate Compensation. The system multiplies frames, doubling or tripling their output to keep the display within an effective dynamic window.
- DisplayPort Adaptive-Sync Protocol. The VESA standard defines Adaptive-Sync as an additional data transmission protocol over the DisplayPort interface. Support is included in the specification starting from version 1.2a. The monitor sends EDID with information about the variable range boundaries, allowing the video driver to control the vertical blanking signal, stretching or shrinking it on demand.
- HDMI Forum VRR Mechanism. The HDMI 2.1 specification implements a native protocol requiring end-to-end support from the cable, source and sink. Unlike DisplayPort-based solutions, this method does not depend on VESA adaptation. It uses fast frame completion signaling and allows TVs to switch display mode on a per-second basis without restarting the video stream.
- G-Sync Hardware Implementation. NVIDIA modular technology involves installing a specialized coprocessor based on FPGA or ASIC into the monitor. The chip takes over panel timing management, replacing the standard scaler. It communicates with the GPU via a proprietary protocol over DisplayPort, providing an extremely wide frequency range and dynamic overdrive adjustment to eliminate ghosting.
- G-Sync Compatible Software Level. This class of devices does not contain a proprietary hardware module but uses the VESA Adaptive-Sync standard. NVIDIA validates production models for the absence of flickering, blanking and artifacts. If the verdict is positive, the driver automatically enables VRR when a certified display is connected, guaranteeing basic quality without a proprietary controller.
- AMD FreeSync Technology. The solution is based on the open VESA Adaptive-Sync standard, requiring no licensing fees from display manufacturers. FreeSync uses a direct connection between the GPU and scaler via DisplayPort or HDMI. Certification levels (Premium, Premium Pro) define mandatory support for LFC, minimum frequency threshold and allowable color deviation at low brightness.
- Buffering Architecture and Triple Buffering. VRR changes the frame buffer filling strategy. The application does not block when the queue overflows but implements an asynchronous update. When the frame rate approaches the display maximum, a built-in frame limiter forcibly holds the render below the hardware V-Sync trigger threshold, preventing a sharp increase in latency.
- Stutter Prevention. Uneven rendering time with a fixed scan causes visual tempo fluctuations. Adaptive frequency smooths this unevenness by synchronizing the physical pixel emission with the mathematical moment of frame completion. The eye perceives motion as continuous because the interval between static image changes becomes congruent with GPU content generation.
- Handling Outliers Outside the VRR Window. When the framerate exceeds the maximum frequency boundary, it activates behavior like Fast Sync or Enhanced Sync depending on the vendor. The driver discards excess frames, keeping the last completed one for output. This prevents entering the V-Sync ON penalty mode, maintaining low latency at the cost of possible micro-roughness at extremely high FPS.
- Power Consumption and Bus Idling. In intervals when a new frame does not arrive and the display frequency is at the lower limit, the GPU memory controller can lower the P-State frequency. This reduces heat and power consumption. The PCIe interface idles, waiting for command buffer activation, which is especially important for battery-powered mobile platforms.
- Interaction with HDR Rendering. When VRR and high dynamic range are activated simultaneously, the monitor must align tone mapping with floating backlight dimming intervals. FreeSync Premium Pro and G-Sync Ultimate regulate the local dimming response outside a fixed PWM grid, avoiding conflict between low-frequency updates and the backlight flicker phase.
- Transformation of Fixed Cinematic Material. Video content with a rigid frequency of 24 or 23.976 frames per second aligns poorly with variable refresh. Media players in a VRR environment forcibly switch the panel to a fixed mode with integer frame multiplication, or the driver applies reverse pull-down to synchronize the presentation render without panning judder.
- The Role of the Driver and Frame Scheduler (Flip Queue). The video driver manages the buffer flip queue. With VRR, the scheduler seeks to minimize pre-rendered frames, keeping the value around one, so that the moment of a key press gets into the currently processed flip as quickly as possible. Queue optimization is critical for esports disciplines.
- Diagnostics and Status Indicators. Activation of adaptive sync is accompanied by visual markers in the monitor on-screen menu. The real-time refresh rate is displayed by a hardware counter that demonstrates hertz fluctuations corresponding to the framerate. If the value freezes at the maximum, the system has exited the VRR window into the fixed V-Sync area.
- Conflict with Desktop Window Manager. In the Windows environment, desktop compositing uses a flip model buffer. For VRR to work in a window, the application must use independent flips via MPO or a full-screen mode based on direct output. Otherwise, DWM buffers frames on a fixed timer, blocking adaptive scan control at the system level.
- The Future of Discrete Variable Ranges. Advanced TV media processors implement dynamic overdrive and subpixel response time compensation linked to the current frequency. This eliminates inverse ghosting that occurs when switching from maximum to minimum refresh rates, making crystal transparency transitions uniform across the entire multiplicative VRR spectrum.
Comparisons
- VRR vs V-Sync. VRR technology dynamically adjusts the display refresh rate to the variable GPU framerate, eliminating screen tearing without locking fps. Vertical synchronization, on the contrary, forcibly binds output to a fixed frequency (60 Hz), which during performance drops causes sharp input lag spikes and unpleasant image judder due to frame repeats.
- VRR vs Fast Sync. Fast Sync allows the GPU to render frames without limitation, displaying only the last completed frame at the next display refresh cycle. This eliminates tearing at high frame rates, but VRR fundamentally surpasses this method during unstable performance, providing perfect smoothness without the micro-delays caused by Fast Sync discarding excess frames.
- VRR vs Enhanced Sync. The Enhanced Sync function from AMD, like Fast Sync, prevents tearing at fps above the monitor frequency but switches to disabled sync when falling below the threshold. The transition between modes can cause noticeable artifacts. VRR operates continuously throughout the supported frequency range, providing seamless tear elimination without abrupt image processing algorithm switches.
- VRR vs Motion Blur Reduction (ULMB/DyAc). Technologies like ULMB or DyAc synchronize backlight strobing with a fixed refresh rate to enhance moving object clarity. Unlike VRR, which eliminates dynamic unevenness, they require a consistently high frame rate and in most implementations cannot work simultaneously with variable refresh rate, forcing the user to choose between smoothness and clarity.
- VRR vs Low Framerate Compensation (LFC). When the frame rate drops below the minimum VRR threshold (for example, 48 Hz), LFC activates frame multiplication, forcing the display to operate at a multiple frequency. This is not a standalone alternative but a compensatory mechanism within the VRR ecosystem, ensuring the absence of tearing and stuttering even at extremely low performance, which is unattainable with classic synchronization methods.
OS and driver support
Implementing VRR at the operating system level requires a mechanism for direct interaction between the graphics driver and the display compositor (Desktop Window Manager), bypassing the vertical sync buffer. In Windows, the WDDM interface version 2.7 and higher, through the flip sequence model (MPO), allows the driver to intercept scan rate control based on VESA Adaptive-Sync extensions; in the Linux environment, the compositor (Wayland or Xorg with patches) uses the DRM atomic modesetting protocol for dynamic VBLANK parameter changes. The driver analyzes the rendering time of each frame (present timing) and sends a signal to the display to extend or shorten the refresh interval, ensuring seamless synchronization without screen tearing.
Security and side-channel attacks
Variable refresh rate creates a potential vector for information leakage through electromagnetic emissions (TEMPEST) and acoustic analysis, since the changing modulation frequency of the pixel matrix generates characteristic radio frequency noise correlating with screen content. Implementing protection requires the introduction of stochastic dithering algorithms into the driver, adding microscopic random delays before sending the frame, which destroys the statistical predictability of the intervals. At the hardware level, the TCON scaler block filters sharp frequency drops, limiting the emission spectrum through smooth interpolation of transitions between hertz values to exclude signal demodulation by an external receiver.
Logging and metrics
The VRR telemetry infrastructure is based on the integration of in-graphics Performance Counters in the driver, which record precise timestamps of frame ready and scanout complete events with microsecond resolution. The graphics API (DirectX or Vulkan), through extensions like DXGI_FRAME_STATISTICS, provides applications access to the flip queue log, calculating the delta between the moment of the Present() call and the actual start of the scan on the display. The operating system aggregates this data in the system scheduler, identifying jitter anomalies, and stores it in a WPA ring buffer for post-mortem analysis by the developer of micro-freeze causes.
Hardware limitations and LFC
When the frame rate drops below the minimum panel scan threshold (usually 40 to 48 Hz), the Low Framerate Compensation (LFC) mechanism, implemented at the level of the graphics card ASIC logic, comes into play. The essence of the operation lies in the multiplication of the last rendered buffer: the display controller forcibly outputs the same frame two or three times in a row at the maximum allowable scan rate, keeping the refresh frequency within the panel operating window and preventing flickering or liquid crystal burn-in. The critical limitation here is the scaler buffer memory capacity: if the difference between the real FPS and its multiple value conflicts with the subsampling buffer, a color shift artifact (overshoot) occurs at object boundaries.
Evolution of standards
Historically, the technology started with proprietary G-Sync modules using a hardware FPGA module in the monitor to capture scan control, which required licensing fees and a complex handshake scheme via Embedded DisplayPort. The transition to the open Adaptive-Sync standard from VESA democratized the technology, integrating control into standard eDP and DisplayPort 1.2a blocks, where the variable frequency is encoded in Secondary Data packets. Modern iterations of HDMI 2.1 VRR and MediaTek VRR in smart displays implement advanced scenario-based frequency switching protocols with Quick Frame Transport, allowing the scaler to rebuild sync pulses during the vertical blanking interval without delay for PLL loop reacquisition.