Dragonwing™ IQ-8275 Evaluation Kit (EVK)
Buy via Distributors
The Qualcomm Dragonwing™ IQ-8275 EVK is a platform to evaluate the Dragonwing IQ-8275 processor for IoT solutions. The EVK is designed to let you choose Yocto Linux or Ubuntu and begin developing. Use Qualcomm multimedia and robotics SDKs to develop robotics and automation projects.
Overview
The Dragonwing IQ-8275 EVK brings the Dragonwing IQ-8275 processor to your bench, with an 8-core heterogeneous CPU, dedicated GPU and NPU, 40 dense INT8 TOPS, and a broad set of I/O and connectivity options. Supporting Yocto Linux or Ubuntu, Qualcomm® multimedia,
and robotics SDKs, it is designed to help accelerate development of robotics, edge AI devices, gateways, autonomous mobile robots, and industrial automation. The kit is designed to support evaluation and prototyping of intelligent edge solutions.
- Your OS, your choice: Canonical Ubuntu, or build lean with upstream Linux and Yocto.
- GenAI at the edge: run Llama2 model of up to 13 billion parameters on-device at around 9 tokens per second, depending on model and configuration.
- Built for vision: Connect up to 12 Cameras with decode/encode capabilities of 4K@135fts/4K@85ftps
- Expandable: a PCIe x4 slot, an M.2 E-key, 2.5GbE and a range of low- and high-speed I/O.
- Open toolchain: bring ONNX Runtime, TensorFlow Lite or PyTorch models, optimized through the Qualcomm® AI Hub.
Edge AI performance with power-conscious design
The Dragonwing IQ-8275 balances compute, AI and graphics on one heterogeneous platform. Workloads run on whichever engine fits them best, designed to balance performance and power use. This can matter for battery-powered robots, fanless gateways and devices that run continuously.
See more. Sense more.
Vision is a key edge AI workload, and the Dragonwing IQ-8275 EVK is designed to support it.
- 3x MIPI CSI (C/D-PHY) camera connectors for multi-camera rigs.
- Concurrent 4K video streams for detection, tracking and inspection.
- Multi-sensor fusion: combine camera data with the onboard IMU and temperature sensor for navigation, monitoring and control.
- Display ready: 2x Mini DisplayPort plus a MIPI DSI connector for HMIs and dashboards.
Designed for robotics and industrial edge AI evaluation
From the first proof of concept toward a production design, the Dragonwing IQ-8275 EVK is designed to support workloads common at the edge:
- AI model development and prototyping: create, test and optimize models for vision, language processing and machine learning.
- Robotics and industrial automation: build proofs of concept for AMRs, drones and UAVs, industrial machinery and agricultural robots, with on-device decision-making.
- Multi-sensor and camera fusion: surveillance, object detection, navigation, predictive maintenance, quality control and process automation.
- Edge computing: industrial IoT, smart cities, retail, worksite safety, smart buildings and healthcare.
Connect it to your peripherals
A broad set of standard interfaces is designed to support the peripherals you already use and the ones you haven't picked yet.
- High-speed: PCIe x4 slot, M.2 E-key, 2.5GbE, USB Type-C.
- Wireless: Qualcomm® Wi-Fi 6E, Bluetooth® 5.3 module on M.2, with two onboard antennas.
- Low-speed: UART, SPI, I²C and 70+ GPIOs for sensors, actuators and custom boards.
- Audio: onboard microphone and speaker.
Ready to get started? Unbox it, flash Ubuntu and begin running your first AI model. The quick start guides, product brief and Qualcomm developer community are designed to help support your path from evaluation to product design.
Find out more in the official documentation:
- Dragonwing IQ-8275 EVK Landing Page
- Quick Start Guide for Ubuntu
-
Quick Start Guide for Linux
-
Dragonwing IQ-8275 SoC Landing Page
- Dragonwing IQ-8275 EVK Product Brief
Dragonwing™ IQ-8275 EVK box contents:
- Dragonwing™ IQ-8275 EVK main board
- Top cover with mini screwdriver
- Power supply with USB Type-C to barrel plug adapter
- USB Type-C cable with Type-A adapter
- Cables
Need help?
Reach out to the Qualcomm Forums for support:
Qualcomm branded products are products of Qualcomm Technologies, Inc. and/or its subsidiaries. Arduino and VENTUNO are trademarks or registered trademarks of Arduino S.r.l.
Tech Specs
| Processor |
Qualcomm Dragonwing™ IQ-8275 (QCS8275-0-AA) |
| Operating systems | Canonical Ubuntu Upstream Linux with Yocto support |
| Memory | 12GB LPDDR5 with link ECC |
| Storage | 128GB UFS microSD card slot |
| Connectivity | 2.5GbE Ethernet Wi-Fi/Bluetooth®: Qualcomm Wi-Fi 6E, Bluetooth 5.3 module, 2x printed antennas |
| Expansion | 1x PCIe x4 slot (switched) 1x M.2 E-key (Wi-Fi or expansion, switched) |
| Camera | 3x MIPI CSI (C/D-PHY) flex connectors |
| Video | 2x Mini DisplayPort 1x MIPI DSI flex connector |
| USB | USB0 Type-C (EDL and fastboot) USB1 Type-C (Host mode) USB2 2.0 (OTG mode, ADB) USB0 and USB1 share one controller and are not active at the same time |
| Interfaces | UART: 2x RX/TX, 1x RX/TX/CTS/RFR, 1x SAIL_UART SPI: 2x SAIL_SPI I²C: 5x I²C, 3x CCI_I2C, 1x SAIL_I2C GPIO: 44x SoC GPIO, 30x PMIC GPIO QUP: 6x QUP (L0/1), 3x QUP (L0–3), 1x QUP (L2/3), 1x QUP (QSPI), 1x QUP (L0–3, CS0, CS1) |
| Audio | 1x microphone, 1x speaker |
| Onboard sensors | IMU: ICM-42688 Temperature sensor: TMP411 |
| Power supply | DC barrel plug (12V-36V) |
| Dimensions / weight | 106 x 106 x 46 mm |
| Operating temperature | -20 to 70 °C |
Get Inspired
FAQs
What is the Dragonwing IQ-8275 EVK for?
It's an evaluation and prototyping platform for the Dragonwing IQ-8275 processor. Use it to test AI workloads, camera pipelines and peripherals before you commit to a custom design. It's designed for evaluating robotics, edge AI devices, gateways and industrial automation use cases.
Which operating systems does it support?
You can choose between Ubuntu, for a familiar desktop experience, and upstream Linux with Yocto, for lean, production-oriented builds. Quick start guides are available for both.
Can it run generative AI and LLMs locally?
Yes. The dedicated NPU is designed to run models of up to 13 billion parameters on-device at around 9 tokens per second, depending on model and configuration. Local processing is designed to help keep data on the device and latency low.
Which AI frameworks can I use?
Bring models from ONNX Runtime, TensorFlow Lite or PyTorch, and optimize them for the Dragonwing IQ-8275 NPU through the Qualcomm® AI Hub.
How many cameras can I connect?
Up to three, using the MIPI CSI (C/D-PHY) flex connectors. The platform supports concurrent 4K video streams for multi-camera vision applications.
What's the difference between the Dragonwing IQ-8275 EVK and the Dragonwing IQ-9075 EVK?
Both are Dragonwing evaluation kits for edge AI and robotics. The Dragonwing IQ-9075 EVK offers up to 100 dense TOPS and 36GB of memory for demanding multi-model workloads. The Dragonwing IQ-8275 EVK offers up to 40 dense TOPS and 12GB of memory, with a balance of performance, power and cost. Check out the Dragonwing IQ-9075 EVK.
Should I choose the Dragonwing IQ-8275 EVK or the Arduino® VENTUNO™ Q board?
Both are built around the Dragonwing IQ8 family. Choose the EVK to evaluate the processor's interface set for a custom hardware design. Choose VENTUNO Q if you want an Arduino board that pairs the processor with a real-time microcontroller, Arduino® App Lab and ready-to-run AI models, designed so you can sense, decide and act on a single board. Discover VENTUNO Q.
Where can I get help?
Check the quick start guides and product brief in the documentation, or join the Qualcomm developer community on Discord and the forum.