GPU Wiki Series
Tegra
NVIDIA system-on-chip family for mobile, embedded, automotive and edge devices. Architecture coverage: Tesla, Fermi, Kepler, Maxwell, Pascal, Volta, Ampere, Blackwell.
Quick Facts
- Architecture
- Ampere, Kepler 2.0, Maxwell 2.0, Pascal, VLIW Vec4, Volta
- Launch range
- 2010-2024
- Memory types
- DDR2, DDR3, DDR3L, LPDDR3, LPDDR4, LPDDR4X, LPDDR5
GPU entries by product stack
Desktop and laptop GPUs are listed in product-stack order.
| GPU | Architecture | Launch Date | Shader Processors | Memory | Bus Width | TDP/TBP | Segment |
|---|---|---|---|---|---|---|---|
| Jetson AGX Orin 32 GB | Ampere | 2023-02-01 | 1,792 | 32 GB LPDDR5 | 256-bit | 40 W | Embedded |
| Jetson AGX Orin 64 GB | Ampere | 2023-03-01 | 2,048 | 64 GB LPDDR5 | 256-bit | 60 W | Embedded |
| Jetson AGX Xavier 16 GB | Volta | 2018-10-01 | 512 | 16 GB LPDDR4X | 256-bit | 30 W | Embedded |
| Jetson AGX Xavier 32 GB | Volta | 2018-10-01 | 512 | 32 GB LPDDR4X | 256-bit | 30 W | Embedded |
| Jetson Nano | Maxwell 2.0 | 2019-03-18 | 128 | 4 GB LPDDR4 | 64-bit | 10 W | Embedded |
| Jetson Orin Nano 4 GB | Ampere | 2023-03-01 | 512 | 4 GB LPDDR5 | 64-bit | 10 W | Embedded |
| Jetson Orin Nano 8 GB | Ampere | 2023-03-01 | 1,024 | 8 GB LPDDR5 | 128-bit | 15 W | Embedded |
| Jetson Orin Nano Super | Ampere | 2024-12-17 | 1,024 | 8 GB LPDDR5 | 128-bit | 25 W | Embedded |
| Jetson Orin NX 8 GB | Ampere | 2023-03-01 | 1,024 | 8 GB LPDDR5 | 128-bit | 20 W | Embedded |
| Jetson Orin NX 16 GB | Ampere | 2023-02-01 | 1,024 | 16 GB LPDDR5 | 128-bit | 25 W | Embedded |
| Jetson TK1 | Kepler 2.0 | 2014-03-25 | 192 | 2 GB DDR3L | 64-bit | 8 W | Embedded |
| Jetson TX1 | Maxwell 2.0 | 2015-11-11 | 256 | 4 GB LPDDR4 | 64-bit | 15 W | Embedded |
| Jetson TX2 | Pascal | 2017-03-07 | 256 | 8 GB LPDDR4 | 128-bit | 15 W | Embedded |
| Jetson Xavier NX 8 GB | Volta | 2020-05-14 | 384 | 8 GB LPDDR4X | 128-bit | 15 W | Embedded |
| Jetson Xavier NX 16 GB | Volta | 2020-05-14 | 384 | 16 GB LPDDR4X | 128-bit | 20 W | Embedded |
| Tegra 2 GPU | VLIW Vec4 | 2010-06-03 | — | 256 MB DDR2 | 32-bit | 20 W | Embedded |
| Tegra 3 GPU | VLIW Vec4 | 2011-11-09 | — | 256 MB DDR3 | 32-bit | 20 W | Embedded |
| Tegra 4 GPU | VLIW Vec4 | 2013-08-01 | — | 512 MB LPDDR3 | 32-bit | 20 W | Embedded |
| Tegra 4i GPU | VLIW Vec4 | 2013-02-19 | — | 512 MB LPDDR3 | 32-bit | 20 W | Embedded |
Series Notes
Tegra put NVIDIA GPU IP into ARM-based SoCs for mobile, embedded and automotive platforms. Software support depends on the device vendor and BSP, not only the SoC.