The Jetson Orin Nano is the major fall GTC 2022 announcement for the Jetson developer community. There’s a new beast in town!
The next generation of the Jetson lineup has been revealed. We already know about the Jetson AGX Orin in 64GB and 32GB flavors, and the Jetson Orin NX modules in 16 and 8GB versions. Todays announcement rounds out the family, the Jetson Orin Nanos in 4GB and 8GB versions.
There’s a full technical blog on NVIDIAs website. Let’s go over some of the highlights. Here’s the architecture diagram:
Here’s the tech specs:
|Jetson Orin Nano 4GB||Jetson Orin Nano 8GB|
|AI Performance||20 Sparse TOPs | 10 Dense TOPs||40 Sparse TOPs | 20 Dense TOPs|
|GPU||512-core NVIDIA Ampere Architecture GPU with 16 Tensor Cores||1024-core NVIDIA Ampere Architecture GPU with 32 Tensor Cores|
|GPU Max Frequency||625 MHz|
|CPU||6-core Arm Cortex-A78AE v8.2 64-bit CPU 1.5 MB L2 + 4 MB L3|
|CPU Max Frequency||1.5 GHz|
|Memory||4GB 64-bit LPDDR5 34 GB/s||8GB 128-bit LPDDR5 68 GB/s|
(Supports external NVMe)
|Video Encode||1080p30 supported by 1-2 CPU cores|
|Video Decode||1x 4K60 (H.265) | 2x 4K30 (H.265) | 5x 1080p60 (H.265) | 11x 1080p30 (H.265)|
|Camera||Up to 4 cameras (8 through virtual channels*) 8 lanes MIPI CSI-2 D-PHY 2.1 (up to 20 Gbps)|
|PCIe||1 x4 + 3 x1 (PCIe Gen3, Root Port, & Endpoint)|
|USB||3x USB 3.2 Gen2 (10 Gbps) 3x USB 2.0|
|Display||1x 4K30 multimode DisplayPort 1.2 (+MST)/e DisplayPort 1.4/HDMI 1.4*|
|Other I/O||3x UART, 2x SPI, 2x I2S, 4x I2C, 1x CAN, DMIC and DSPK, PWM, GPIOs|
|Power||5W – 10W||7W – 15W|
|Mechanical||69.6 mm x 45 mm 260-pin SO-DIMM connector|
Table 1. Jetson Orin Nano series specification – Priced in 1K units
Through the Looking Glass
Looking at the specs, the first thing that jumps out is storage. Earlier Jetsons have either an attached SD card or eMMC memory. The Jetson Orin Nano has neither. Instead the module supports external NVMe. The obvious benefit is that the amount of external storage is configurable and is fast. Most earlier Jetson modules have eMMC which always seems to be too small as applications become more and more advanced (and storage hungry).
The 8GB Orin Nano is about $100 less than a 8GB Orin NX. The 8GB Orin Nano has about 40 Sparse 8 Int TOPS, whereas the Orin NX has 70. The Orin NX appears to get most of this gain from a higher power handling capabilities, 15W vs 25W. Also, the Orin NX has hardware video encoders, the Orin Nano uses two CPU cores instead.
Speaking of the CPU cores, the Orin Nano divides them into two blocks. The first block is 4 CPU cores, the second block is 2 CPU cores. That seems reasonable if there’s some specific use cases like video encoding that require “reserving” a CPU block to complete the task.
Another interesting announcement is the immediate availability of emulators for all members of the Jetson Orin family. That means that using a Jetson AGX Orin Development Kit, you can emulate any of the other Orins, such as the NX or Nano. Get a little head start on your apps, as the Orin Nano modules are due out in January.
This is a very interesting turn of events. I was very much expecting a toned down version of the Xavier as the new Nano. Instead NVIDIA did something much smarter. The Nano? We’ll just make it a slightly toned down version of the latest and greatest. This promises to be great fun!