DX-H1 Quattro
Supercharge Your AI Workloads
with 4x Efficiency
Overview
Features | Details |
|---|---|
AI Performance | 100 TOPS (=400 eTOPS / INT8) for real-time AI inference |
Power Consumption | 20 W / 6V |
Interface | PCIe Gen3 x16 (Bifurcation x4x4x4x4 requires configuration) |
Memory | LPDDR5 16GB (2GB x 8) |
Storage/Operating Humidity | 5% to 90% RH (Non-condensing) |
Operating Temperature | -25℃ to 85℃ (Throttling) |
-25℃ to 65℃ (Non-Throttling) | |
Storage Temperature | -40℃ to 125℃ |
Dimensions | 167 x 68.9 x 7.75 mm |
Form Factor | Low Profile PCIe Card |
Key Benefits
Exceptional Performance and EfficiencyThe DX-H1 Quattro delivers a peak performance of 100 TOPS (INT8). This raw power is matched by a remarkably low Thermal Design Power (TDP) of just 20W, making it a leader in energy efficiency for AI inference.
Innovative Interface DesignThe DX-H1 Quattro connects via a PCIe Gen3 x16 slot. However, its unique design requires the motherboard to support Bifurcation x4x4x4x4. This technology allows the single x16 slot to be split into four separate x4 lanes, enabling each of the card's four internal processors to communicate with the host system.
Broad System CompatibilityThis card is built to integrate seamlessly into diverse computing environments. It supports both x86-based and ARM aarch-64 based CPU architectures. For software, it is compatible with 64-bit Microsoft Windows and various Linux distributions, including Ubuntu.
Compact Form FactorDespite its powerful capabilities, the DX-H1 Quattro is designed as a Low Profile PCIe Card. Its compact dimensions ensure it can be easily integrated into a wide range of systems.
Related Products
Solutions to Simplify AI Adoption and Optimize Surveillance Infrastructure for TCO Reduction.
You can quickly deploy AI video analysis capabilities without completely overhauling your infrastructure
by installing a server equipped with the DX-H1 Quattro (AI Accelerator) to work with your existing CCTV system.
Product Models | Accelerator Configuration | AI Inference Performance (Based YOLOv5, FHD, 25 FPS) | Recommended Specifications | Certified Partners Products |
|---|---|---|---|---|
 Vision AI. 1U AI Workstation |  Up to 1x DX-H1 Quattro | Up to 128 CH. (Max. 3,200 FPS) |
|
|
 Vision AI. 1U AI Server |  Up to 3x DX-H1 Quattro | Up to 384 CH. (Max. 9,600 FPS) |
|
|
 Vision AI. 2U AI Server |  Up to 6x DX-H1 Quattro | Up to 768 CH. (Max. 19,200 FPS) |
|
|
Note: Performance metrics are based on the AI Model YOLOv5s_512 at 25FPS.
Actual channel support may vary depending on the specific AI Model and FPS settings.
Use Cases
Data Center AI Servers
The DX-H1 Quattro is ideal for cloud-based AI services or large-scale AI model inference servers due to its high performance of 100 TOPS (INT8) and its PCIe Gen3 x16 interface.
Multi-Channel Video Analytics Systems
The Bifurcation x4x4x4x4 support makes it an ideal solution for systems that require multiple AI processors to simultaneously handle numerous video streams.
Smart Factories and Smart Cities
The card can be utilized to implement various AI applications by analyzing real-time data from industrial cameras and sensors, such as defect inspection, safety detection, and traffic analysis.
Edge AI Servers and Appliances
With a remarkably low Thermal Design Power (TDP) of just 20W, the card delivers stable AI performance in edge environments where power efficiency is critical.
Network Video Recorders (NVRs)
It can be used in security and surveillance systems to analyze video from multiple channels in real time, performing AI-based object detection and behavior recognition.
Industrial Robots and Automation Equipment
Its high-performance AI inference capabilities can accelerate robot vision systems and efficiently handle complex tasks like autonomous navigation and precision control.