Surely you’ve heard of GPGPU, that is, how to use the GPU for general purposes. Just like the CPU and what it does. This has its advantages and disadvantages, as we are going to see in this article. In addition to learning everything about this type of computing, applications, examples, etc. Homogeneous computing is computing that has been taking place since the beginning of computing, that is, where the CPU was in charge of general processing, leaving other chips with specific tasks. However, currently the term heterogeneous computing has emerg, which breaks with this lack of versatility and flexibility when processing, as you will see in the following section. Heterogeneous computing is increasingly important. Starting with data centers and HPC, which are the first to benefit from it, until now many other consumer products that have also the benefits of this solution where not only the CPU is for general purpose, it can also be use other types of processors.
Examples of heterogeneous architecture implementations
Here are some examples of heterogeneous architectures in some well-known product. AMD’s APU is a magnificent example of this, since it integrates a GPU and a CPU. On the same monolithic chip, sharing memory for both types Japan Phone Number Data of processors. Arm also its heterogeneous architecture big. LITTLE , where it mixes lower-performance Arm Cortex A-Series CPU cores with more compact sizes. As well as more efficient, with other larger, more powerful and higher consumption Cortex A-Series cores . In this way, the workload can be sent, depending on its weight, to one or the other to improve efficiency. This is what is known as heterogeneous multicore. Intel began a few generations ago to do something similar in the x area. It now integrates its P-Core and E-Core, that is, Performance cores and Efficient cores. The first with a more complete ISA in terms of vector extensions, larger in terms of silicon size, with higher performance, but also with higher consumption.
What is GPGPU?
GPGPU (General-Purpose Graphics Processing Unit) stands for general-purpose graphics processing unit. And is a term that refers to a modern heterogeneous computing methodology where the. GPU is to assist the CPU with Albania Phone Number some workloads. Regardless of that these loads are not graphical tasks for which these GPUs were designed. Keep in mind that a GPU is designed to improve parallelism with floating point calculations, and this allows a GPU to achieve greater performance when performing these calculations compared to a CPU. But for that to be possible, the software to be adapte for it, and that involves using an API different from the graphics APIs (DirectX, OpenGL, Vulkan,…). In the case of hardware, absolutely nothing to be modifi. All modern GPUs are GPGPUs , you just have to take advantage of their programmable processors to perform calculations in parallel and massively instead of the CPU having to do it more slowly.
