Why is the hardware not using 100% capacity? Why is my video card idle? Is this processor bottleneck on my video card or not? We will explain in this article how the interaction between video card and processor works, and how such “is not 100%” may not be a clear indicative of who your system’s bottleneck is.
For those who want to investigate this deeply on your PC, we first recommend that you install a real -time performance analysis tool on your machine. We have this tutorial here that shows how to install the MSI afterburner combo and Riva Tuner and how to configure them on your screen. Another option, which is even more efficient for what we are going to do here, is Intel Presentmon.
Intel’s software brings many hardware metrics, as well as the MSI afterburner, but introduces a very interesting concept: the GPU Busy.
Because my hardware is not 100%
The most important point about using your computer’s components and their efficiency is to understand how a picture is made in games. The process begins with the processor, which needs to deal with multiple game -related works, such as physics, Ia and also the drawing of frame design.
It is this last command that ends the processor part, and starts the part of the video card, which is to make all steps to deliver the frame to display on the screen. Different games and different settings make the time CPU and GPU take to do the job vary.
Most of the time, what people want most when setting up their computer is the 100% video card. That’s when many believe they took more from their investment, and if they bought the whole video card, it is to use it all. Below we have an example in graph showing the production time of the frames. Each color (dark blue, white and light blue) represent work on a different picture:
As you can see, the time between each new table coincides with the time the video card takes to produce a new picture. This is because this is a scenario of limitation by the performance of the video card, or as it is usually called in English, a scenario GPU Bound (linked to the GPU, in free translation).
But that does not necessarily mean that this is the ideal scenario. In some games, especially those focused on competitive, it is normal for the graphics to be much simpler. The goal is of course: take light on the game’s look helps it run at high frame rates. And what can happen, that’s it below:
As there is little work for the GPU, she ends her part of the work long before she has a new picture being required by the CPU. The result is the video card being stopped, waiting for this new work, and a use of GPU that is not represented by 100% in statistics.
Your PC will always have a slower component. And sometimes it won’t be the GPU
And this is where despair begins. For some, you have to upgrade the processor. For others, you are putting money away. The reality is simple: it will always have a slower component. Sometimes it will be the CPU, sometimes it will be the GPU.
It’s good to keep an eye on if this video card subuse is not caused by something more serious, such as lack of RAM (what you can correct with a more memory upgrade), lack of video memory (which you can solve reducing the quality of textures) or game optimization problems (there is nothing to do, it is sitting and crying).
If it’s not any of these problems, then the culprit is your processor. And it’s okay. There will be games that will be like this. You can take advantage of this idle video card time and climb the graphic quality more, something that will hardly affect the frame rate until you do the video card’s working time exceed the processor’s working time.
But what if both, the processor and the video card, do not appear 100% of use? Here we have a difference between how the CPU and GPU operate.
Hardware is not 100%: CPU or GPU?
The CPU may already be giving the most of its performance, even though it does not appear as in 100% of use
Graphic chips are made up of many computational nuclei, shadders, extremely capable of distributing part of the work to each. This way you can easily reach 100% chip use.
Processors deal with many work cycles that are not so efficient in parallelism. It has processes that cannot be done in parallel, and need to wait for another part to be done before. Games are notorious not to use all processor threads, with most games not climbing their performance on processors with more than six cores. We talk more about this in this article here.
So with the processor often, we see it present a use below 100%, and yet it is already delivering the maximum performance it can offer. It can not distribute more functions to more processing cores, and is already running the game the fastest way it can. In the chart above it is noticeable as from the 6 nuclei, Counter Strike 2 performance reaches its apex, and does not keep climbing more even when we enable much more nuclei.
Our conclusion here is similar to the video about bottleneck. You don’t have to be neurotic with hardware use in each game, after all, each one will depend more on the performance of one or the other, sometimes both. What matters is that both deliver your work at 16 milliseconds or less in games you are aiming at 60fps in the gameplay, and about 6 milliseconds or less in competitive games for you to have high frame rates. No one needs to be 100% use.
Source: https://www.adrenaline.com.br/tutorial/nao-ta-usando-tudo-quando-seu-hardware-nao-esta-batendo-100/
