In e-sports games, it may only take less than a second for the master to pass the move to determine the winner or loser, for example, in the "PUBG" corner to meet love, who shot first, hit first, who can get a huge advantage in the confrontation, and even directly determine the winner and loser, decide who takes the next flight first. Therefore, hardware manufacturers can be said to spare no effort in reducing game latency for e-sports players, such as NVIDIA Reflex and AMD Anti-Lag, which are technologies born to reduce system latency, which allows players to get a more streamlined control experience in the game.
But we've also received feedback from some of our classmates that the role of NVIDIA Reflex and AMD Anti-Lag hasn't been evident by a game accelerator yet. This is obviously a confusion between the concepts of system latency and network latency, but it also leads us to an exploration idea, which one has a greater impact on the game experience for PC games, system latency or network latency?
So what is system latency?
For PC, system latency mainly refers to the computer's local computing time, for example, in FPS games, from the player clicking the mouse to the display finally shows the corresponding change in the process, the time it takes, is what we often call the system delay, which can be subdivided into several different components, we can simply understand from the following figure.
In general, system latency can be divided into three parts: peripherals, computer internals, and display latency. The latency of peripherals is well understood, taking the mouse as an example, part of the mouse receives the delay of the player's click signal, and the other part is the time when the mouse sends the signal, connected to the mouse target USB interface to process these signals; the internal delay of the computer is divided into data transmission time and computing rendering time, in the PC game, the former includes sampling, submission rendering, driving, etc., mainly the CPU is responsible for processing, the latter is responsible for the graphics card, including the rendering sequence and the actual rendering The display delay refers to the fact that before the graphics card renders the picture to the actual display of the frame on the display, it is also necessary to scan the output and the monitor to process the picture before it can finally be output to the player's eyes, which also takes a certain amount of time to complete.
So what is network latency?
For local stand-alone games, network latency can be regarded as a non-existent concept, but for online games, the impact of network latency is very obvious. Network latency in online games, generally speaking, is that you make an operation, the corresponding data is transmitted to the server through the network, and then the server responds and calculates, and the corresponding result is passed back to you for the time required.
In this data transmission process, most of them are node-to-node, and there may be cross-region, cross-operator situations, so the final feedback to us is much higher than the network latency time of the system delay, sometimes up to hundreds of milliseconds.
Which has the greater impact on system latency or network latency?
System latency: Optimizations on the software lead to smooth operation
In fact, this problem does not exist for stand-alone games, because stand-alone games do not need to be networked, and the quality of the network does not affect the game experience. Therefore, this issue is basically only worth discussing for online games. We've done a comparison of NVIDIA Reflex and AMD Anti-Lag in improving system latency, so let's review the test results at the time. Let's take Overwatch as an example and look at the system latency of this game at 4K resolution.
We can see that in the "Overwatch" in 4K resolution, the native system latency of gtX 1060 graphics cards and RX 590 GME graphics cards has reached a level of more than 100 milliseconds, and the system latency corresponding to RTX 3060 and RX 6600 graphics cards is lower, which is at the level of 80 milliseconds, from here we can see that upgrading the hardware can indeed bring lower system latency, and the experience in the actual game is that the frame rate is higher and the operation is smoother.
After turning on NVIDIA Reflex and AMD Anti-Lag, we can see that the former is better than the latter, which can bring 20 milliseconds to 25 milliseconds of optimization, and the latter is only about 10 milliseconds of improvement, but overall there is still an improvement. Compared to the fundamental changes brought about by the upgrade of hardware, turning on NVIDIA Reflex and AMD Anti-Lag has little impact on the number of frames in the game, but the smoothness of operation will change, but this change may require esports-level sensitivity to experience.
Network latency: Very important for overseas service games
So how much does network latency affect connected gaming? In fact, we have also tested this in PUBG PUBG, played 5 consecutive PUBG games, and recorded the network latency value displayed by the game after each lap of the first three hoops, and then took the corresponding maximum value as the final result, and made the following table.
The comparison in the table is actually the network latency brought by the wired network and the wireless network in the actual online game, from the effect point of view, the wired network is indeed the best, but the performance of the wireless network is not too bad, and it is generally a few milliseconds higher. In contrast, the network latency caused by the on or off of accelerators is very amazing, can reach the level of 200 milliseconds, not counting the problem of packet loss, and the impact on the game experience is very large. Therefore, for online games, system latency and network latency are not a "two-or" problem, but a matter to be solved in both aspects.
summary:
For stand-alone games, since there is no need for networking and no server to perform data calculations, all of its latency comes from system latency. The most fundamental way to shorten system latency is to improve the performance of hardware, but this often means a relatively large capital investment. In contrast, system latency improvements brought about by software optimizations such as NVIDIA Reflex or AMD Anti-Lag may be more cost-effective, especially for older hardware that is already less performant.
However, to say that this NVIDIA Reflex and AMD Anti-Lag that is more useful, from the performance point of view, the former is superior, can easily bring tens of milliseconds of optimization, but it needs application support, not necessarily effective in any game; the latter is the driving level of the function, after enabling all applications will benefit, but the effect is limited, sometimes only a few milliseconds of optimization, far less than the former.
For online games, both system latency and network latency will affect the player's gaming experience, but who is more important? If the network latency itself has been controlled within a reasonable range, it may bring more benefits from improving the system latency; but if the network latency has reached the level of hundreds of milliseconds, then there is no doubt that you need to solve the network latency problem, and the difference in system latency is not worth mentioning in the face of this huge network latency.
Of course, whether it is system latency or network latency, improving them is more just to reduce the disadvantages of your game, especially in competitive games, and plays a role in the icing on the cake. But if you need the effect of "charcoal in the snow", for example, if you are facing problems such as the average frame rate of the game is not up to standard, the network latency reaches hundreds of milliseconds, and the operation has a visible time difference, you need not only "improve" but "solve", upgrade the hardware, use game accelerators, etc. is the real cure.