Different Architectures of NVidia GPUs

In today's blog, we will be discussing the different types of architecture(current) of the NVidia GPUs. You might have noticed some terms like Pascal, Turing and Ampere. These are the three types of architecture in which NVidia Graphics Card works.

Pascal Architecture:

Pascal is a codename for the GPU architecture developed by NVidia. This architecture was introduced in 2016. This architecture is primarily used in the GeForce 10 Series starting with the GeForce GTX 1080 and GTX 1070, which were released on May 17, 2016, and June 10, 2016, respectively. The architecture was named after the French mathematician Blaise Pascal. This architecture has 64-128 CUDA cores, a memory bandwidth of 720GiB/s, and NVlink. They also announced the RayTracing support for Pascal Architecture in 2019 after a driver update,

Turing Architecture:

Turing is a codename for the GPU architecture developed by NVidia. This architecture was introduced in 2018. This architecture was primarily used by RTX 20 Series. This architecture followed the same naming convention as before. This architecture was named after the computer scientist Alan Turing. This architecture introduced RayTracing Technology - an efficient way to calculate many light rays. In February 2019, NVidia released GTX 16 series with Turing Architecture, but it lacked CUDA cores when compared to other RTX GPUs. It uses RT cores  Tensor(AI) cores to calculate lighting besides CUDA. 

Ampere Architecture:

.We have already talked about this in our previous blogs. Please do check that too. This Ampere Technology was recently introduced in September. This architecture has twice Tensor Cores and RT cores than its predecessor Turing. This architecture was developed with Samsung. This architecture was named after the French mathematician and physicist André-Marie Ampère