Graphics Processing Unit

Graphics processing units, which process large amounts of data in parallel, have become the foundational hardware for AI systems that rely on hardware capable of processing millions of text parameters simultaneously.

CUDA, the computing platform that transformed graphics processing units from pixel renderers into general-purpose components powering AI, originated from a computer science student.

The history of graphics processing units began with hardware for displaying visuals in arcade machines, but the transition to 3D graphics led to the microchips and software that power today's AI data centers.

The term GPU was first used by Sony in 1994 to refer to the geometry processing unit inside the PlayStation console, five years before Nvidia's marketing of the GeForce 256 revised the acronym to graphics processing unit.

Watch a tech demo of the world's first consumer graphics processing unit, the Nvidia GeForce 256, released on October 11, 1999.

Although GPU stands for graphics processing unit, the acronym is often used to refer to a graphics card—an assembly that houses the microchip along with supporting components.

When buying a custom computer, selecting integrated graphics is more affordable, but a dedicated graphics card should be considered for CAD, video content creation, high-end gaming, and 3D modeling.

Explore graphics processing unit benchmarks for 2026, led in performance by Nvidia's GeForce RTX 5090, with AMD's Radeon RX 9060 XT 8GB and Nvidia's GeForce RTX 5060 providing the best 1080p performance per dollar.

Take a virtual tour inside the NVIDIA GA102 graphics processing unit to explore the architecture that enables it to perform over 35 trillion calculations per second.

With many smaller computing cores, graphics processing units process many identical tasks in parallel, whereas central processing units have fewer, larger cores built to handle a wide variety of more complex tasks sequentially.