Definition Of GPU – Features, Specifications, Size & More

Definition Of GPU

GPU is the acronym for Graphics Processing Unit and represents precisely the heart of a graphics card, just like the CPU does in a PC. Apart from the heart, your brain is also in charge of carrying out all the complex calculations that allow us to enjoy our games on the screen. Usually it is the piece of silicon that had both AMD and NVIDIA or Intel manufacture( where the transistors are recorded). To put it commonly, it is the so-called graphics card chip. This concept is important from the point of view of utility since, as we have mentioned, a GPU is not a graphics card in itself.

On the other hand, a graphics card does integrate a GPU under its belt. In addition, it needs a PCB, resistors, SMDs, capacitors, voltage drivers, phase regulators, VRAM, display outputs, and more.

Main Features And Specifications

A GPU is based on a silicon chip that is a series of millions of transistors under a specific architecture and has dedicated features for each model. The importance of the concept itself comes to us from APUs (for example) where they also house a GPU, only they do so on a much smaller substrate and hand in hand with a CPU.

The  Technical characteristics of a GPU depend on the specific model; it depends on the particular chip, so we can only generalize when talking about the concept.

The Average Size of a GPU

The size of GPU is much larger than a CPU die, and nowadays, it isn’t easy to get them below 100 mm2, even in the most advanced lithographic processes. It includes a substrate size that is generally between three and four times the area of ​​the chip, mainly due to the number of elements connected to it. It is a very basic interposer to wire connections with the chip . GPU connects the bus, the VRAM, the control signals, voltages, and clocks with our protagonist.

The most advanced GPU processes is a vertical scaling called 2.5D IC. It is still a separate unit and should be treated as such. Currently the design is being questioned since soon we will be unable to talk about one GPU per graphics card.

The Different Elements of a GPU

These are some additional elements of GPU

• Command processor: Reads the screen list or instructions sent by the CPU to generate graphics and do complex calculations.
• Rasterization Unit: It performs the transformation from a three-dimensional space based on vertices to a two-dimensional space based on pixels.
• Texturing Unit: It is responsible for applying an image on a surface to simulate texture or color in real life.
• Raster Output: The unit drawing the final pixels on the image buffer. It is the only piece with the last level Cache that has to write permissions on the VRAM.
• Shader Unit: A kernel that executes programs to manipulate graphics primitives in real time.
• The intersection unit Calculates the intersection of the rays of the scene with the objects. It is essential for Ray Tracing.
• Tessellation unit: The unit subdivides the vertices of objects to give them a more rounded and polished appearance.
• Video CODEC: Independent processor that decodes video in various multimedia formats and reproduces them.
• Memory interface: What allows the GPU to read from its RAM, known as VRAM.
• DMA: Unit that allows the GPU to read from the main RAM of the system.

Conclusion

Basically, more elements are added to a GPU, such as NVIDIA RTX graphics, where specific parts have also been added for ray tracing processing of RT Cores stc.

Also Read – About 5G Network –Definition, Benefits & More