These types of technologies working in symbiotic mode are going to be common from now on, since the goal of both Intel and its rival AMD is to offer complete processing ecosystems. Obviously things like memory will be left out, but these days GPUs aren’t just used to display pretty graphics in games. There are many applications that use the so-called GPU computing pipeline for support.
When we talk about the computing pipeline, we are referring to the use of the graphics processor, not to generate images on the screen. That is, the dead times in the GPU are used so that it helps the CPU to solve certain problems faster and in some environments even where the GPU does not generate images as is the case with high performance computing.
Dynamic PowerShare
It is a feature designed for gaming laptops that consists of the Intel Core CPU and the ARC GPU sharing the same power control circuitry, so it is still the Intel equivalent of AMD SmartShift. What allows this? Easy, when the CPU workload is low and therefore it is consuming less power than normal this allows the GPU to have a small increase in its clock speed by taking advantage of the excess power left by the CPU and vice versa.
The unit in charge of controlling the temperature, clock speed and voltage takes into account the parameters of both processors to see when it can speed up the other, but it is not limited to that. It also takes into account the energy cost of future instructions, allowing it to “predict” when to perform the corresponding uploads and downloads. To all this we must take into account the ability to “learn” and, therefore, to automate certain common tasks related to certain applications.
We must clarify that the Dynamic Power Share does not mean that consumption is always static and distributed, since in low-load scenarios the CPU and GPU clock speed will be reduced to increase battery life and the time between charges. .
The iGPU will no longer be a hindrance in Deep Link
Normally, when we connect a dedicated graphics card it ceases to be useful and becomes a piece of hardware that when, for example, we play a game, or we are using an application like Blender, it ceases to be useful. Intel’s idea with Deep Link is that the iGPU is not just for decoration and that it serves to at least speed up computing tasks in conjunction with ARC graphics. Which will mean that those who use Intel Core processors without this element will not be able to take advantage of this technology.
This ability is reminiscent of the Cell Broadband Engine, the PS3 CPU that brought a series of cores very similar to the GPU Compute Units that were used to speed up certain tasks. For example, for eimage post-processing effects or for perform the preculling, which consists of rendering the image without textures or shader programs to locate the superfluous geometry in the scene. With this it is achieved that everything that we do not see is not calculated by the GPU and to be able to obtain more frames per second.
However it is not the only application that we can see, a developer can even use Deep Link to configure the iGPU as a advanced sound card and offer us audio at the level of high-end dedicated hardware. The applications are many, but it will depend on programmers optimizing for a purely Intel ecosystem.
Artificial intelligence combined in Deep Link
The most popular and clearest improvement that artificial intelligence brings in image processing is super resolution. Which consists of using deep learning prediction algorithms to generate a version with more pixels and, therefore, that has less resolution from one that has less. Intel has implemented the algorithm Gigapixel AI of Topaz inside Deep Link to generate images at higher resolution using the iGPU of the Intel Core for it.
It is a functionality that people who work with images on a daily basis and need to obtain them at a higher resolution will appreciate. Since it will allow us to do it without depending on external remote services, directly from our PC and with the advantage of having both GPUs combined to generate the highest quality version.
hyperencode
The use of Deep Link is not only limited to computing, but also to hardware video codecs. Which means that those that come with the iGPU will be able to work in tandem with those of the graphics card in order to share the work and finish encoding or decoding images from one video format and resolution to another in less time. It’s a capability that if you’re working on programs like Adobe Premiere, SONY Vegas or you simply dedicate yourself to real-time streaming on platforms like Twitch or YouTube, you’ll appreciate it in terms of performance.
