| Architecture | Release Date | Compute Capability | GeForce | Quadro | Jetson |
|---|---|---|---|---|---|
| Fermi | 2010 | 2.0 2.1 |
GeForce 400 series GeForce 500 series |
Quadro 600 | |
| Kepler | 2012 | 3.0 3.5 3.7 |
GeForce 600 series GeForce 700 series |
Quadro K600 | |
| Maxwell | 2014 | 5.0 5.2 5.3 |
GeForce 900 series | Quadro K620 | Jetson Nano |
| Pascal | 2016 | 6.0 6.1 6.2 |
GeForce 10 series | Quadro P600 | Jetson TX2 |
| Volta | 2017 | 7.0 7.2 |
Nvidia TITAN V | Quadro GV100 | Jetson Xavier NX Jetson AGX Xavier |
| Turing | 2018 | 7.5 | GeForce 20 series | NVIDIA T600 Quadro RTX 4000 |
|
| Ampere | 2020 | 8.0 8.6 8.7 |
GeForce 30 series | NVIDIA RTX A2000 | Jetson Orin NX Jetson AGX Orin |
| Ada Lovelace | 2022 | 8.9 | GeForce 40 series | NVIDIA RTX 2000 | |
| Hopper | 2022 | 9.0 | H100(for Data Center) | ||
| Blackwell | 2024 | 10.x | GeForce 50 series |
When you use --generate-code option, Compute Capability is expressed as 2 digit number.
CC 5.0 is --generate-code arch=compute_50,code=sm_50
Reference:
Full list of products and Compute Capability
https://developer.nvidia.com/cuda-gpus
Compute Capability Support
https://docs.nvidia.com/deploy/cuda-compatibility/index.html#support-hardware__table-hardware-support
Technical Specifications per Compute Capability
https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#compute-capabilities