NVIDIA Jetson Orin Nano Super Developer Kit is an upgrade to the Jetson Orin Nano Developer Kit with 1.7 times more generative AI performance, a 70% increase in performance to 67 INT8 TOPS, and about half the price, making it a great development platform for generative AI at the edge, mostly robotics.
We’ve seen several AI boxes and boards in the last year capable of offline generative AI applications like the Firefly AIO-1684XQ motherboard or Radxa Fogwise Airbox which I reviewed with Llama3, Stable diffusion, Imgsearch, etc… A product like the Fogwise Airbox delivers up to 32 TOPS (INT8) and sells for around $330 which was very competitive then (June 2024). However, the Jetson Orin Nano Super Developer Kit will certainly disrupt the market with over twice the performance, a lower price, and a larger developer community.
NVIDIA Jetson Orin Nano Super specifications:
- NVIDIA Jetson Orin Nano 8GB Module
- CPU – 6-core Arm Cortex-A78AE v8.2 64-bit CPU @ 1.7 GHz with 1.5MB L2 + 4MB L3
- GPU – 1,020 MHz NVIDIA Ampere architecture with 1024 CUDA cores and 32 tensor cores
- 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)
- Memory – 8GB 128-bit LPDDR5 (102GB/s)
- Storage – Interfaces for SD card and NVMe SSD on carrier board
- AI Performance – 67 TOPS (Sparse), 33 TOPS (Dense)
- Power Modes – 7W, 15, or 25W
- Storage
- microSD slot
- 2x M.2 socket for NVMe SSD (See Expansion section)
- Display interface – DisplayPort 1.2 (+MST) connector
- Camera interfaces – 2x 22-pin MIPI CSI-2 camera connectors
- Networking
- Gigabit Ethernet RJ45 port
- Optional WiFi and Bluetooth via M.2 Key-E slot
- USB
- 4x USB 3.2 Gen2 Type-A ports
- USB Type-C port for UFP (Upstream Facing Port)
- Expansion
- M.2 Key M slot with PCIe Gen3 x4
- M.2 Key M slot with PCIe Gen3 x2
- M.2 Key E slot
- 40-pin GPIO expansion header with UART, SPI, I2S, I2C, etc…
- Misc
- 12-pin button header
- 4-pin fan header
- Power Supply – DC power jack
- Dimensions – 103 x 90.5 x 34.77mm (with feet, carrier board, module, and thermal solution)
The highlights in bold show the kit is basically the same as the original Jetson Nano Orin Developer Kit. However, CPU frequency is now boosted to 1.7 GHz (instead of 1.5 GHz), the GPU runs at 1,050 MHz (vs 635 MHz), the memory bandwidth is quite higher (from 68GB/s to 102GB/s) and they added a 25W power mode, all of which explain the extra performance. The dimensions are different too, but maybe that’s just because they measured it differently.
What’s most surprising is that NVIDIA says all those performance improvements are just software updates, so the original NVIDIA devkit can also benefit from the 1.7x boost in generative AI performance for LLM (Large Language Model), LVM (Large Vision Model), and ViTs (Vision Transformers) models.
The extra performance will be available to NVIDIA Jetson Orin NX and Nano module with the Nano 4GB reaching 34 TOPS, the Nano 8GB 67 TOPS, the NX 8GB 117 TOPS, and the NX 16GB 157 TOPS. To update the software you need to get JetPack 6.1 with support for boosted performance using the SDK Manager (select JetPack 6.1 (rev. 1)) and change the power mode to MAXN as follows:
|
1 |
sudo nvpmodel -m 2 |
Software support remains the same with the NVIDIA AI software stack including the NVIDIA Isaac for robotics, NVIDIA Metropolis for vision AI, and NVIDIA Holoscan for sensor processing. The company also claims developers can save significant time with the NVIDIA Omniverse Replicator for synthetic data generation (SDG) and NVIDIA TAO Toolkit for fine-tuning pre-trained AI models from the NVIDIA NGC catalog.
If you don’t already own an NVIDIA Jetson Orin Nano, you can purchase the “Super” model for $249 on Seeed Studio, Arrow, or Sparkfun provided it’s still in stock. NVIDIA did not mention pricing for the modules, e.g. for mass production, but I doubt they also slashed the prices in the same fashion as for the devkit. The NVIDIA Jetson Orin Nano is probably a sponsored product for developers. Additional information can be found on the product page.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
What is the max ram that this can use? Is it like the M2 M3 macs that use shared memory?
The RAM is soldered on the module, so it’s just 8GB unless the Jetson module is replaced with one with 16GB.
Yeah that’s what I noticed as well. It’s a bit sad, because in this price range, doubling the RAM size or even quadrupling it wouldn’t have significantly inflated the price and would make it much more suitable for larger models. From the demos I’ve seen, the performance is quite decent, so the board would be usable with larger models with more RAM.