Frame rate has been the only metric for comparing performance in games. Even at smooth frame rates, however, gaming can be uncomfortable if there is a significant amount of input lag – the game responds to controls with noticeable lag. There are countless tools for measuring frame rates, but unfortunately not for testing responsiveness. Here we introduce the specialized Frame Latency Meter recently made available by AMD.
In single-player games, players typically prefer the highest possible graphics quality. In competitive gaming, the opposite approach is more common. There are several reasons for this: lowering graphical details helps achieve higher frame rates (FPS), which reduces latency. It also removes unnecessary visual effects and objects, simplifying the scene and making it easier to spot important elements and stay oriented.
With higher frame rates, motion becomes less choppy and more fluid, allowing players to better track fast-moving objects and react more precisely to sudden scene changes. The reduction in significant differences between frames makes characters’ motion easier to recognize and makes aiming easier. Incoming shots are also more distinct, allowing players to anticipate and dodge them more effectively.
Lower latency is crucial for reducing the delay between a player’s input, like pressing a key or clicking a mouse, and seeing the action taking effect on the screen. Minimizing this lag allows players to react more quickly to on-screen events, resulting in gameplay that feels more responsive and seamless.
The average human reaction time to visual signals is typically between 200 and 250 milliseconds. For experienced players who have trained their reflexes, this time can be slightly shorter, usually between 150 and 200 milliseconds. Given the reaction times being so long, one might think that an additional difference of forty or twenty milliseconds in a game’s response time wouldn’t matter much, as human senses can’t detect such a slight variation.
In real life, responses are instantaneous, which is what the human brain expects. Any further delay disrupts this expectation, making players feel they have less precise control over their actions. Although a difference of 20ms might be too subtle to consciously notice, it significantly impacts the brain’s ability to synchronize inputs with on-screen responses.
A long response time causes your character or vehicle to react in a delayed manner to your input, making the controls feel slow and unresponsive. This makes quick turns, dodging attacks, or precise aiming more difficult, often causing players to react later to what’s happening on the screen.
With practice, you can learn to compensate for the delay by moving the mouse or pressing keys in advance. However, during rapid scene changes, it increases the probability of failing to react appropriately.
You’ll notice this when you don’t hit doorways perfectly while running, start turning either too early or too late in a corner while driving, mistime a jump, or have your shots go wildly off target instead of achieving a headshot.
That’s why it’s important not to focus solely on high frame rates. Response times are also crucial during the process of optimizing your system and game settings, as they significantly impact your gameplay experience.
There are many tools for measuring frame rate. You can even use an application connected to AMD, Nvidia or Intel drivers to display the current frame rate in an overlay text. Measuring response time is much more complicated.
The basic principle is simple; just record the time of a button press or mouse movement and measure how long it takes for it to show up on the screen. It’s also easy to measure if you have the necessary equipment – a high-speed camera or a latency measurement tool like Nvidia LDAT or the open-source OSLTT. With these tools, you can measure the entire response from mouse input to the displayed action on the monitor – the system latency.
Source: Nvidia.com
Response time can also be measured software way. The peripheral latency and display latency are not reflected in the overall response time, but it is possible to measure how long it takes to process and render the image on the computer itself (PC latency).
To do this, Nvidia is using Reflex libraries that are integrated into games and are able to detect mouse action and when the graphics card has rendered a frame. The problem is that they only work on GeForce cards and in games with Reflex integrated.
It’s probably not easy to develop a universal application – you can capture mouse movement, but you can’t easily find out how much time the game spent rendering the image and exact moment it displays updated the image.
Nevertheless, AMD recently introduced an alternative to both specialized tools and Nvidia’s Reflex – an open source Frame Latency Meter application. It neatly sidesteps the problem of the game’s inability to look into the render pipeline by using an API commonly used for image recording to detect the change. It works on all manufacturers’ graphics chips, supports DirectX 11 and DirectX 12 APIs, and Windows 10 and Windows 11 operating systems.
It uses Advanced Media Framework (AMF) for screen capture on AMD cards, and Desktop Duplication (DXGI) on other cards. With DXGI, it can then run on any graphics card. The output is latency statistics per frame in a CSV file.
What principle it works on and what it can do, we will discuss further…
