rCore-Q9075
- Features
- Product Details
- Block Diagram
- Interfaces
- Hardware Spec
- Design Resources
- Documentations
- Accessories
Product Details
The rCore-Q9075 is an edge computing module based on the Qualcomm IQ9075 processor. It delivers high computing power and exceptional energy efficiency, enabling stable handling of heavy workloads even in extreme environments while providing unparalleled on-device AI performance for demanding industrial applications.
Block Diagram
Interfaces
Hardware Spec
SoC
Qualcomm Dragonwing™ IQ-9075
CPU
Octa-Core Kryo Gen 6 (Cortex-A78C based) @2.36 GHz
GPU
Adreno 663 GPU
- Delivers 1.2 TFLOPS FP32 with secure GPU compute
- Supports Vulkan 1.2, OpenGL ES 3.2, OpenCL 2.0 FP
- Supports Adreno NN Direct
MCU
Quad-Core Cortex-R52 @ 1.85GHz
NPU
200 TOPS@INT8 (Sparse Computing Power)
- Features the Hexagon Tensor Processor (HTP) with quad HVX and dual HMX
- Supports deep-learning frameworks: TensorFlow, PyTorch, ONNX, Paddle, Caffe, DarkNet, etc.
RAM
36GB 96-bit LPDDR5@6400MT/s (with ECC support)
Audio DSP (LPASS)
1980 MPPS, 7x TDM/I2S
3x High-Speed I2S for Radio FE
Display Support
Max 48 Mpix and 12 displays with no superframe
5x 4K60, 3x 4K60 + 8x 1080p60, 2x DSI + 2x DP/eDP MST2 + 2x DP/eDP MST4
Video Decode
Formats: AV1, H.264, H.265, VP9, MPEG2
1x 8K60, 2x 8K30, 4x 4K60, 8x 4K30, 16x 1080p60, 32x 1080p30
Video Encode
Formats: H.264, H.265, HEIF/ HEIC
2x 4K60, 4x 4K30, 8x 1080p60, 16x 1080p30
Camera
Up to 16 cameras, max 12 MP sensor resolution , 4x 4-lane CSI2
PCIe
2x PCIe Gen 4: 1x 2-lane + 1x 4-lane (Root Complex & Endpoint)
USB
2x USB 3.1 Gen 2, 1x USB 2.0
Networking
2x 2.5 GbE w/ TSN (SGMII)
Other I/O
21x QUP_SEs (supports UART/I2C/SPI), 149x GPIOs
Storage
2x UFS 3.1 Gen 4 2-lane, 1x 8-bit SDCC5, NVMe over PCIe
WiFi/BT/WAN
Support through companion chips: QCA6698AQ , SDX35 , SDX72
Operating Conditions
-40°C to 115°C (Tj)
Software
Linux Yocto, Linux Ubuntu
Dimension
65 mm × 65 mm
Design Resources
Carrier Board
Radxa rCore-Q9075 IO Board
Contact us for more details and to request reference design files.
Documentations
Category
ALL
FAQs
IQ-9075
Adreno 663
200 TOPS
PCIe Gen4
Size: 65x65 mm
Ethernet
Display 4K
LPDDR5
Powered by Qualcomm Dragonwing IQ-9075
Octa-core Kryo Gen6 CPU up to 2.36 GHz, Adreno 663 GPU, and dual Hexagon NPUs delivering up to 200 TOPS — combining CPU, GPU, and NPU acceleration for edge AI.
200 TOPS
IQ9 Series Platform
- Kryo Gen 6 CPU
- Adreno 663 GPU
- Hexagon Tensor Processor with vector and matrix accelerators (Al/NPU)
- LPDDR5 Memory With ECC
- Qualcomm® Spectra™ 690 ISP, CV Processors, multi-camera HDR
- 3x Audio DSPs with Al accelerator
- Dual Adreno 1199 Display Processing Unit (DPU)
- IO USB3.x, PCle Gen 4 Ethernet with TSN support UART, I2S, I2C, SPI
- Adreno Video Processing Unit (VPU) 675 Gen 5,
- MCU-like subsystem with dedicated real-time cores, high-speed IOs and RTOS
Flagship Video Engine
Multi-Stream UHD, Real-Time Processing
Support:
- AV1 / HEVC / H.265 / H.264 / VP9 / MPEG-2 video decoder up to 1x 8Kp60, 4x 4Kp60 or 32x 1080p30
- H.264 / H.265 / HEIF / HEIC video encoder up to 2x 4Kp60 or 16x 1080p30
- Concurrent 2x 4Kp60 encode and 2x 4Kp60 decode
Comprehensive and Rich AI Ecosystem
Support Yocto or Ubuntu with TensorFlow, PyTorch, ONNX, and more.
Run GPT, LLaMA, and Stable Diffusion locally — secure and low-latency.
Features the Hexagon Tensor Processor (HTP)
with quad HVX and dual HMX
Support Yocto or Ubuntu and ready for AI deployment.
Extensive Compatibility
Compatible with major frameworks: TensorFlow, PyTorch, ONNX, Paddle, Caffe, DarkNet, etc.
Ideal For Local AI APPs
Run popular open-source apps with ease
High-Performance Memory & Versatile Storage
Optimized for high bandwidth and flexible storage expansion
- Six-channel 3200 MHz LPDDR5
SDRAM + 3 MB system cache - Dedicated internal memory (256 kB
IMEM, 1.5 MB GMEM, etc.) - Dual UFS 3.1 + eMMC 5.1 + multi-
interface support (PCIe, NVMe, etc.)
Hardware-Grade Security Protection
Hardware RNG + ICE protection
- Hardware ECC/RSA encryption
- 5th-gen CE5 Engine
- FIPS/CAVP Certified
Size & Configurable GPIOs
Main Domain GPIO: 149 ports, RTSS Domain GPIO: 78 ports
- Octa-Core Kryo Gen 6
@2.36 GHz - Adreno 663 GPU
@840 MHz - Quad-core Cortex-R52
real-time subsystem
@1.85 GHz
Applications & Use Cases
Fogwise AIRbox Q900 comes with integrated rCore-Q9075, powering its core AI edge computing performance
Edge Generative AI
Deploy private GPT or Stable Diffusion models directly on-site—instant, confidential, offline.
Real-time Computer Vision
Multi-camera analytics for surveillance, inspection, smart buildings, or robotics.
Industrial Automation
On-device logic for industrial control, predictive maintenance, sensor fusion.
Remote and Harsh Environments
Rugged hardware suited for factory floors, retail automation, smart healthcare kiosks.
Education & Research
Features high AI computing power, versatile interfaces, and AI ecosystems to support prototyping and algorithm validation.
Industrial Gateways
Enables multi-protocol data aggregation, real-time edge AI processing, and secure industrial data transmission as a robust core.
Comprehensive expansion capabilities
Audio
LS-I2S (muxed with PCM/TDM pins)
HS-I2S (muxed with PCM/TDM pins)
PCM / PDM (muxed with LS-I2S and HS I2S pins)
Display
4x eDP / DP
2x MIPI DSI 4-lane
Camera
4x MIPI CSI 4-lane
Storage
2x UFS 3.1 Gear 4 2-lane
1x eMMC 5.1 (via 8‑bit SDIO)
NVMe (via PCIe)
USB
2x USB 3.1 Gen2 HOST and OTG Port
1x USB 2.0 HOST and OTG Port
Ethernet
2x 2.5 GbE w/ TSN (SGMII)
Connectivity
1x PCIe Gen4 x2
1x PCIe Gen4 x4
UART
QSPI
SPI
I2C
Comprehensive Technical Documentation and
Active Community
Comprehensive Technical Documentation and Active Community !
Radxa provides comprehensive technical documentation, empowering users to fully utilize their products. At the same time, they actively maintain an engaged community, fostering communication and collaboration among users.
Attention:
The terms HDMI, HDMI High-Definition Multimedia Interface, HDMI Trade dress and the HDMI Logos are trademarks or registered trademarks of HDMI Licensing Administrator, Inc.
Radxa rCore-Q9075
High-Computing Power, High-Energy Efficiency Module
Radxa rCore-Q9075
This product is exclusively available through our network of Approved Partners. To find out more about our trusted partner network, please visit their websites.