The world's most powerful AI computer built for energy-efficient autonomous machines powers your next-generation product. Compared to the previous generation, it can achieve 6 times the performance and 200 TOPS of computing power for multiple concurrent AI inference pipelines, and high-speed interface support for multiple sensors, and is suitable for applications in manufacturing, logistics, retail and healthcare ideal solution。
Built for the next generation of AI products
- size
- performance
- power
Compact form factor with improved performance
Measuring 100mm x 87mm, the Jetson AGX Orin™ offers up to 6 times the performance in the same compact form factor compared to the Jetson AGX Xavier. This makes it the solution of choice for the new generation of autonomous machines such as distribution and logistics robots, factory systems and large industrial UAVs.
High performance for autonomous machines
Jetson AGX Orin features NVIDIA Ampere architecture GPU and Arm® Cortex®-A78AE CPU, as well as next-generation deep learning and vision accelerators. High-speed IO, 204GB/s of memory bandwidth, and 32GB of DRAM can feed multiple concurrent AI application pipelines. With Jetson AGX Orin, you can now deploy larger, more complex models addressing problems such as natural language understanding, 3D perception, and multi-sensor fusion.
Select power consumption
Jetson AGX Orin provides higher performance for autonomous machines in an energy efficient system. The module consumes only 15 watts, capped at 50 watts, and is form factor and pin compatible with the Jetson AGX Xavier, while delivering up to 6x the performance. NVIDIA Jetson AGX Orin with NVIDIA JetPack™ and NVIDIA AI Software Stack is your next-generation AI product platform.
Jetson AGX Orin Module Specifications
Jetson AGX Orin 32GB | Jetson AGX Orin 64GB | Jetson AGX Orin 開(kāi)發(fà)者(zhě)套(tào)件(jiàn) | |
| AI 性(xìng)能 | 200 TOPS | 275 TOPS | |
| GPU | 搭载(zài) 56 个(gè) Tensor Core 的(de) 1792 核 NVIDIA Ampere 架构 GPU | 搭载(zài) 64 个(gè) Tensor Core 的(de) 2048 核 NVIDIA Ampere 架构 GPU | |
| GPU 最大(dà)頻率 | 930 MHz | 1.3 GHz | |
| CPU | 8 核 Arm® Cortex®-A78AE v8.2 64 位(wèi) CPU 2MB L2 + 4MB L3 | 12 核 Arm® Cortex®-A78AE v8.2 64 位(wèi) CPU 3MB L2 + 6MB L3 | |
| CPU 最大(dà)頻率 | 2.2 GHz | ||
| DL 加速器 | 2x NVDLA v2 | ||
| DLA 最大(dà)頻率 | 1.4 GHz | 1.6 GHz | |
| 視覺加速器 | 1x PVA v2 | ||
| 安全(quán)集群(qún)引擎 | - | ||
| 顯存 | 32GB 256 位(wèi) LPDDR5 | 64 GB 256 位(wèi) LPDDR5 | 32GB 256 位(wèi) LPDDR5 204.8GB/s |
| 存儲 | 64GB eMMC 5.1 | ||
| 視頻編码 | 1x 4K60 (H.265) 3x 4K30 (H.265) 6x 1080p60 (H.265) 12x 1080p30 (H.265) | 2x 4K60 (H.265) 4x 4K30 (H.265) 8x 1080p60 (H.265) 16x 1080p30 (H.265) | |
| 視頻解(jiě)码 | 1x 8K30 (H.265) 2x 4K60 (H.265) 4x 4K30 (H.265) 9x 1080p60 (H.265) 18x 1080p30 (H.265) | 1x 8K30 (H.265) 3x 4K60 (H.265) 7x 4K30 (H.265) 11x 1080p60 (H.265) 22x 1080p30 (H.265) | |
| 攝像头(tóu) | 多(duō)达(dá) 6 个(gè)攝像头(tóu)（通(tòng)过(guò)虛拟通(tòng)道(dào)最多(duō)可(kě)支持(chí) 16 个(gè)） 16 通(tòng)道(dào) MIPI CSI-2 D-PHY 2.1（高(gāo)达(dá) 40Gbps）| C-PHY 2.0（高(gāo)达(dá) 164Gbps） | 16 通(tòng)道(dào) MIPI CSI-2 連(lián)接器 | |
| PCIe* | 高(gāo)达(dá) 2 x8 + 1 x4 + 2 x1 （PCIe 4.0，根(gēn)端口(kǒu)和(hé)端點(diǎn)） | 支持(chí) x8 PCIe 4.0 的(de) x16 PCIe 插槽 支持(chí) x4 PCIe 4.0 的(de) M.2 Key M 插槽 支持(chí) x1 PCIe 4.0 的(de) M.2 Key E 插槽 | |
| USB* | 3x USB 3.2 2.0 (10 Gbps) 4x USB 2.0 | USB Type-C 連(lián)接器：2x USB 3.2 2.0 USB Type-A 連(lián)接器：2x USB 3.2 2.0，2x USB 3.2 1.0 USB Micro-B 連(lián)接器：USB 2.0 | |
| 网(wǎng)络* | 1x GbE 1x 10GbE | RJ45 連(lián)接器，至高(gāo)可(kě)支持(chí) 10 GbE | |
| 顯示接口(kǒu) | 1x 8K60 多(duō)模 DP 1.4a (+MST)/eDP 1.4a/HDMI 2.1 | 1x DisplayPort 1.4a (+MST) 連(lián)接器 | |
| 其他(tā) I/O | 4x UART、3x SPI、4x I2S、8x I2C、2x CAN、PWM、DMIC 和(hé) DSPK、GPIO | 40 針(zhēn)接头(tóu)（UART、SPI、I2S、I2C、CAN、PWM、DMIC、GPIO） 12 針(zhēn)自(zì)動(dòng)化(huà)接头(tóu) 10 針(zhēn)音(yīn)頻面(miàn)闆接头(tóu) 10 針(zhēn) JTAG 接头(tóu) 4 針(zhēn)風(fēng)扇(shàn)接头(tóu) 2 針(zhēn) RTC 電(diàn)池備份連(lián)接器 microSD 插槽 直(zhí)流電(diàn)源插座 電(diàn)源、強(qiáng)制恢複和(hé)複位(wèi)按鈕 | |
| 功耗 | 15 瓦(wǎ)至 40 瓦(wǎ) | 15 瓦(wǎ) – 60 瓦(wǎ) | |
| 規格尺(chǐ)寸(cùn) | 100 毫(háo)米(mǐ) x 87 毫(háo)米(mǐ) 699 引腳(jiǎo) Molex Mirror Mezz 連(lián)接器 集成(chéng)导热(rè)闆 | 110 毫(háo)米(mǐ) x 110 毫(háo)米(mǐ) x 71.65 毫(háo)米(mǐ) （高(gāo)度(dù)包括支架、载(zài)闆、模組和(hé)散(sàn)热(rè)解(jiě)決方(fāng)案(àn)） | |
Learn more about Jetson AGX Orin。
Jetson AGX Orin Developer Kit
The NVIDIA Jetson AGX Orin Developer Kit contains the power-efficient, high-performance Jetson AGX Orin module. Featuring up to 200 TOPS of computing power runn...
NVIDIA JetPack SDK
The NVIDIA JetPack SDK provides access to all Jetson hardware capabilities and allows developers to apply the world-class NVIDIA AI software stack to their next...
pretrained model
Many AI applications face common needs: classification, object detection, language translation, text-to-speech, recommender system engines, sentiment analysis, ...
NVIDIA Platforms Built for Robotics, Healthcare and Smart Cities
NVIDIA software platforms including Isaac™ for robotics, Clara™ for healthcare, and Metropolis for smart cities accelerate vertical-specific applications。
NVIDIA Isaac is built for robotics
Robots create new efficiencies and improve quality of life in industries such as manufacturing, logistics, healthcare and services. NVIDIA Isaac accelerates thi...
NVIDIA Clara is built for healthcare and life sciences
NVIDIA Clara is a healthcare application framework for the development and deployment of AI-powered imaging, genomics, and smart sensors. It includes full-stack...
NVIDIA Metropolis is built for smart cities
NVIDIA Metropolis includes an application framework, a set of developer tools, and an ecosystem of partners that combine visual data and AI to improve operation...