He PC hardware for gaming It is used rather little, since there are components within the circuitry that are not used at all and that if they are used, the performance of the games for all users will be increased at zero cost, that is, free of charge. Why does this happen and what benefits would there be if it were used?
The advances in recent years regarding the development of PC games should have put an end to the myth that it is a platform where optimizing games to make better use of the system was something for dedicated systems. However, most programmers decide to ignore a large part of the system.
Multithreading and works on PC hardware for gaming
In games, the extra cores and execution threads are used to execute certain complementary tasks to the main game process, but which are part of it and are essential as they are complementary to it. The methodology is usually to have a general process that acts as a conductor for all the others and distribute the rest between the different cores of the main CPU.
More or less the process works as follows:
- The main game process divides its work into several subtasks, where each of them is similar in behavior, but refers to different objects in the game. This allows you to stop work and be able to process these subtasks in blocks, which reduces the time the CPU takes on each frame.
- The delegation of the subtasks is organized by jobs, each job depends on a processor core or, failing that, on the processor execution thread. Once this is done, each of the jobs is performed in parallel.
The idea is simple, a single core would take much longer to do the work than several at the same time, it is still the classic division of labor and it helps to reduce the time per frame that the processor occupies, however, it is not everything is as beautiful as it seems.
Windows always wants its share and the more the merrier
Windows is like a feudal lord who wants his share and, therefore, will also use the processor’s resources for his things, which is why we recommend not always having background tasks. What’s more, it’s so selfish that when it comes to distributing the jobs among the different execution threads, it doesn’t do it efficiently. The problem comes from the fact that a multi-threaded kernel is not equivalent in power to two cores, but what is done is to take advantage of inactivity bubbles to take advantage of that performance.
Of course we should not confuse performance with power, one is the percentage of utilization of the work capacity and the other is the work capacity itself. They are totally different concepts. The fact is that when a game asks what resources the processor has, it doesn’t lie, but the operating system takes the opportunity to take part of those resources.
On consoles, on the other hand, one or two cores are usually reserved for the operating system to work in the background, but this is something simple for a system to play, not a complete computer where the operating system will be competing for resources in a way continued.
What is not used in PC hardware to play
However, in gaming PC hardware there are a number of components that are not usually taken into account when developing games, for example, GPUs are not only good for displaying graphics, but also for calculating the physics of the same and, therefore, part of their power can be used to achieve these calculations. For the graphics card it represents a much lower percentage of work capacity usage than the CPU.
However, it is not the only case, for example we have audio codecs on motherboards, network cards and other elements that could be used to reduce the impact on the CPU. Simply by assigning specialized jobs to them and freeing up the central processor cores. Another case is DMA units, which can be used to move data between memories without the participation of the central processor in an easier and more efficient way.
What we mean is that almost no developer uses these resources on PC and they are readily available thanks to the DirectX API. Apparently, it is much easier to use the central processor than not to take advantage of the resources already available on the users’ computers.
