What exactly does a graphics card do?
We all know that a Graphics Card is used to play games and produce visual output, but I want to dive deeper into the working of a graphics card.A Processing unit, be it a CPU (Central Processing Unit) or a GPU (Graphical Processing Unit) both perform mathematical operations on binary (0 and 1).You mave have heard of cores? The cores are basically ALUs (Arithmetic Logic Unit) and perform arithmetic operations.
Read also:All you need to know about GRAPHIC PROCESSING UNIT
So, a CPU has small number of cores (as compared to a GPU) usually 2 or 4 cores at present. (AMD has manufactured CPUs with 64 cores, 3990X) These cores are very powerful on their own and each of them are assigned different tasks to perform (called multitasking). The main emphasis in CPU is low latency, i.e. user can interact with different processes in the real time.
A GPU, on the other hand, has a significant number of cores.A GTX 1050Ti is a low-end GPU with 768 cores (CUDA Cores; we won’t go into what CUDA stands for for now).The cores in a GPU are not as powerful as the cores in a CPU, but because of the large number of cores, a mathematical calculation can be completed rapidly since a GPU employs all of its cores to work on a single activity, whereas a CPU focuses on multitasking.The primary goal of a GPU is to boost throughput (giving more output per unit time).
Millions of pixels make up an image.The CPU delivers data about each pixel to the GPU, which then uses it to execute complicated mathematical and geometrical calculations.The GPU requires a place to store information and completed images as it develops them.It does this by storing data about each pixel, its color, and its placement on the screen in the card’s RAM (called VRAM).
A portion of the VRAM can also serve as a frame buffer, storing completed images until they are ready to be displayed. Video RAM often runs at fast speeds and is dual ported, which means the system can read and write to it at the same time.When it’s time to show the image, the GPU generates it and displays it on the screen. To summarize, processing an image necessitates a large number of mathematical and geometrical computations, and a GPU is specialized hardware designed to perform these calculations quickly.
Scientific research, data analysis, and CAD modeling all involve mathematical calculations, hence GPUs are used in these areas as well. You can game on CPU but it simply won’t be able to render images quickly enough and you will encounter what is called lag, stutter, low FPS etc.
Also the actual chip is called a GPU, whereas the chip assembled on a motherboard, along with VRAM, heatsink and I/O is called a graphics card.
What are graphic cards?
A graphics card is a piece of hardware that is installed into your computer with the sole purpose of processing images.The cards are plugged into the motherboard and require extensive power to operate. A common 400w power supply in a generic desktop will not be enough for a dedicated video card. Its not uncommon to see one require a few hundred watts for itself, though actual power usage has become far more efficient as transistor sizes have shrunk, and technology has advanced.
Read also:Difference between GRAPHICS PROCESSING UNIT(GPU) and VIDEO GRAPHICS ARRAY(VGA)
They utilize dedicated memory (ram) and their own processor (gpu) to handle the computation without involving the rest of the system resources.The cards are specifically designed for high performance. While your CPU may be a quad core processor, the GPU on your graphics card can have upwards of 1000 cores. This extensive power is what allows the system to play high resolution games when the system’s resources would be otherwise unable.
Graphics cards are needed for any type of real gaming. Below is a high end card geared toward gaming, but they have an alternative use in graphic design, video rendering, and mathematical computation. Because they are designed for parallel computing, they handle mathematics extremely efficiently compared to a conventional processor. I often use my cards for MATLAB, though I bought them to game.
Graphics Card i should avoid
RX550, RX560, GTX 1050, GTX 1050Ti & GTX 1650 due to their current prices. In 2019, I’d highly recommend getting at least an RX570, which sometimes goes for as low as $110 during sales. It is more powerful than all the above listed GPUs & is somewhat futureproof as well, should run games for the next 3 years or so comfortably.I would say avoid GTX 1060 at $200 or so as well because RX580 performs similar to it & costs under $180, or you can buy GTX 1660 for as low as $210.
Moving upwards, I feel Vega 64 isn’t a good deal at $400 either. With upcoming price cuts for RTX GPUs rumored, even the new Navi GPUs might be a bad value at launch price.