Graphics processing unit

Nvidia GeForce 3 GPU (introduced in 2001)

A graphics processing unit (GPU) is a processor that renders (or creates) images, animations, graphics and then displays them on the computer’s screen. A strong GPU is able to process complex animations and graphics smoothly and efficiently.

Images which GPUs calculate (render) may be 2D or 3D. For most applications, the GPU has to render two dimensional (2D) pictures. Modern GPUs are so powerful, that there is no difference in 2D performance between low-cost and expensive GPUs.

Some applications use 3D pictures to simulate three-dimensionality. Examples include video games, and applications for design and technical construction (CAD). The 3D acceleration performance of GPUs differs greatly. Generally, more expensive GPUs can render faster than low-cost ones.

Origin

The first graphics cards were used in the 1970’s, within arcade machines, as a cheaper alternative to Random-Access Memory (RAM). However, graphics cards were not viable for consumer products until the release of the Large Scale Integration (LSI) circuit chip in the 1980’s. During the early-to-mid 1990’s, graphics cards that were capable of 3D support became more common in arcade, computer and console games. The first consumer-level GPU, meant for personal computers, was the Nvidia GeForce 256, which was released in 1999.

Modern Uses

Nowadays, GPUs are omnipresent in personal computers, mobile telephones and video game systems as most CPUs come with an internal GPU built-in. Those internal GPUs are good enough to render the many visual effects of modern graphical user interfaces such as shadows, semi-transparent backgrounds and animations. Often, they even render the graphics for older video games. However, for newer video games a more powerful additional dedicated GPU is usually required.

Other devices such as virtual reality headsets and driverless cars, such as Tesla, also use GPUs. The tasks a GPU can do include technical calculations, Artificial Intelligence problem solving, photo editing and other applications.

GPU Computing

The term GPU computing is also known under the term GPGPU. It combines all tasks a GPU possibly can do. This is far more than geometric calculations and the output of pictures. Tasks are calculation of physics, Artificial Intelligence or even acceleration of video and picture editing. One of the first applications to support GPU computing was Adobe’s Photoshop.

NVidia’s approach: CUDA

NVidia calls their GPU computing framwork CUDA. It is very popular but can be used only with GPUs from NVidia. CUDA is based on the C programming language.

OpenCL

OpenCL is a vendor-independent, general-purpose GPU computing framwork, introduced in 2008. It runs on all major GPUs, also on those from Nvidia.

External GPU

An external GPU is a graphics processor located outside of the housing of the computer. External graphics processors are sometimes used with laptop computers which lack a powerful graphics processor. Therefore, attaching a GPU to some notebook can be desirable.

GPU Accelerated Video Decoding

Some GPUs have special hardware to decode and encode video streams. This can greatly reduce the load on the CPU which saves power and also allows the CPU to do other work. As of 2023, the video decoding standards H.264 (also called AVC), H265 (also called HEVC), VP9 and AV1 are widely used.

GPU Interfaces

Early GPUs used the Peripheral Component Interconnect (PCI) interface to communicate with the motherboard of a computer. As graphics cards became more powerful, a faster kind of interface called Advanced Graphics Port (AGP) was created. Finally, from 2006 onwards, the PCI Express (PCIe) standard was used. It provides for up to 16 physical express lanes per card slot. The PCIe standard was frequently updated and is still the current overall standard.

PCI graphics cards will only work in PCI slots; AGP will only work in AGP slots. PCIe graphics cards will function only in PCIe slots. However, some older graphics cards will not work in every motherboard, even if placed in the correct slot.

Multi-GPU systems

Multi-GPU Systems are computers which use more than one GPU. Generally, this was used in high-end home computers to accelerate computer games. However nowadays, most games do not support this method anymore.

But there is still the ability to have one GPU rendering normal game scenes and one GPU calculating physics. This is currently supported by nVidia’s GPUs only and called PhysX. ATi and Intel have developed their own physics engine named Havok.

History of Multi-GPU systems

The first attempts in Multi-GPU Systems were done by now defunct company 3Dfx in the late 1990s, which was bought by nVidia later. Their Voodoo video cards, which were 3D only accelerators, could be hooked up to a second Voodoo video card to gain more performance. This was called SLi. Theoretically the performance would double, practically it increased far less, depending on the video game. Later, 3Dfx built even four GPUs on one unreleased video card.