What Are CUDA Cores in Video Cards?

CUDA Cores Explained

CUDA stands for Compute Unified Device Architecture. It's a technology developed by NVIDIA that accelerates GPU (graphics processing units) computation processes.

With CUDA, researchers and software developers can send C, C++, and Fortran code directly to the GPU without using assembly code. This lets them take advantage of parallel computing in which thousands of tasks, or threads, can be executed simultaneously.

More Information on CUDA Cores

You may have seen the term CUDA when shopping for an NVIDIA video card. If you look at the packaging of such a card, or read video card reviews, you'll often see a reference to its number of CUDA cores.

CUDA cores are parallel processors similar to a processor in a computer, which may be a dual or quad-core processor. NVIDIA GPU's, however, may have several thousand cores. These cores are responsible for various tasks which allow the number of cores to directly relate to the speed and power of the GPU.

Since CUDA cores are responsible for dealing with all the data that moves through a GPU, the cores are what handle things like the graphics in video games, for situations such as when characters and scenery are loading.

Applications can be built to take advantage of the increased performance offered by CUDA cores. You can see a list of these applications on NVIDIA's GPU Applications page.

If you need device drivers for your CUDA video card, see our guide on How to Update Drivers in Windows.

CUDA cores are similar to AMD's Stream Processors; they're just named differently. However, you can not equate a, say, 300 CUDA NVIDIA GPU with a 300 Stream Processor AMD GPU.

For more information on CUDA, visit NVIDIA's About CUDA page.

Choosing a Video Card With CUDA

In short, a higher number of CUDA cores typically means that the video card provides faster performance overall. However, the number of CUDA cores is only one of several things to consider when choosing a video card.

NVIDIA offers a range of cards that feature as few as 8 CUDA cores, like in the GeForce G100, to as many as 5,760 CUDA cores in the GeForce GTX TITAN Z.

Graphics cards that have Tesla, Fermi, Kepler, Maxwell, and Pascal architecture support CUDA.