AIEH4000
PCIe x16 AI Accelerator Card
Hailo-8™ AI Processor x4
-
Compliant with PCI Express 3.0 x16
-
Supports Single PCIe Slot with low-profile Form-Factor
-
Powered by Hailo-8™ AI inference processors.
-
Supports 104 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Note: AIEH4000 is incompatible with AMD Platform when using x16 Link Width.
- Compliant with PCI Express 3.0 x16
- Supports Single PCIe Slot with low-profile Form-Factor
- Powered by Hailo-8™ AI inference processors.
- Supports 104 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Introduction
SUNIX PCIe x16 AI Acceleration Card, AIEH4000, enables Hailo-8™ AI Processor into Edge computers, provides an easily deployable solution for AI developers to offload CPU loading on low-latency deep learning inference and cost-effective Edge AI computing with higher processing capabilities and lower power consumption. And this card can deliver up to 104 TOPS, enabling it to handle the most demanding computer vision applications. [Test Results: ResNet-50 v1( 224x224)@5,328FPS, YOLOv5m (640x640)@872FPS]
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
Features
SUNIX AIEH2000 VS NVIDIA A4000 AI visual computing acceleration performance comparison test(YOLOV5M)
SUNIX's AI accelerator card. This card not only significantly reduces computer energy consumption but also demonstrates exceptional performance in high-speed AI visual recognition computations. The AI accelerator card, AIEH2000, introduced by SUNIX, is a PCIe interface AI accelerator card specifically designed for PC. To showcase the performance and energy-saving capabilities of the SUNIX AI accelerator card, we conducted tests using the same testing platform as the Nvidia RTX A4000 graphics acceleration card, which already incorporates AI model inference capabilities. Employing the yolov5m model for image recognition tests allowed us to compare the performance of SUNIX's AIEH2000 with Nvidia A4000 and understand their differences in performance. The test results displayed on the screen vividly illustrate the data for both accelerator cards: SUNIX AIEH2000 achieves an average processing speed of approximately 400 frames per second (FPS), capable of handling up to 400 real-time frames per second. Meanwhile, the Nvidia A4000 processes approximately 300 frames per second (FPS), indicating its capability to handle around 300 frames per second. These test results indicate that utilizing SUNIX's AIEH2000 AI accelerator card for image recognition processing delivers a speed approximately 100 frames per second (FPS) faster than the Nvidia A4000 graphics card, equating to a computational speed increase of about 1.3 times. Regarding energy consumption, SUNIX's AIEH2000 demonstrates exceptional performance. During model inference computations, the AIEH2000 consumes an average of only 10 watts, while the Nvidia A4000 consumes approximately 60-100 watts. Based on testing and calculations, SUNIX Tech's AIEH2000 achieves a computation speed of 27 FPS per watt, compared to Nvidia A4000's mere 3 FPS per watt. Therefore, SUNIX's AI accelerator card not only excels in efficiency but also exhibits clear advantages in energy conservation and cost-effectiveness. Especially when considering the economic and environmental aspects of deploying AI edge computing frameworks on a large scale, the differences between the two become more pronounced.Specifications
Board Description |
|
| Model | AIEH4000 |
| Description | PCIe x16 AI Acceleration Card |
| AI Processors | Hailo-8™ AI Processor x4 |
| AI Performance | 104 TOPs (Tera-Operations Per Second) |
| Power Consumption | 27W (based on model) |
| PCIe Interface | PCI Express Gen3 x16 |
| AI Frameworks | TensorFLow, TensorFlow Lite, ONNX, Keras, Pytorch |
| Hailo Software Suites |
Dataflow Compiler (Model conversion and compilation) HailoRT (Runtime environment and driver) Model Zoo (Pre-trained models) TAPPAS (Deployment framework, examples and multi-network pipelines) |
| OS Support | Linux (e.g. Ubuntu, Yocto) & Windows 10 (X86/X64) / 11 |
| Certification | CE/FCC/VCCI/BSMI |
| Green Environment | RoHS |
Environment |
|
| Operation Temperature | 0 to 50°C (32 to 122°F) commercial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -20 to 85°C (-4 to 185°F) |
Dimension |
|
| PCB Dimension | 167.0 x 68.5 mm |
| Bracket | Standard 121 mm / Low profile 80mm(Optional) |
Support
Download Datasheet
| Datasheet | AIEH4000 Datasheet EN |
|
Download Guide
| Guide | AI Acceleration Card Quick Installation Guide |
|
Download Driver
| Driver | HailoRT - Windows | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:39:48 | |
| Size | 8.7MB | |
| OS |
Windows 10 64bit
Windows 11 64bit
|
| Driver | HailoRT - Linux - x86_64 | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:38:53 | |
| Size | 6.2MB | |
| OS |
Linux Kernel 5.x
|
| Driver | HailoRT PCIe Driver - Linux | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:37:31 | |
| Size | 132.8KB | |
| OS |
Linux Kernel 5.x
|
AIEH4000
PCIe x16 AI Accelerator Card
Hailo-8™ AI Processor x4
-
Compliant with PCI Express 3.0 x16
-
Supports Single PCIe Slot with low-profile Form-Factor
-
Powered by Hailo-8™ AI inference processors.
-
Supports 104 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Note: AIEH4000 is incompatible with AMD Platform when using x16 Link Width.
- Compliant with PCI Express 3.0 x16
- Supports Single PCIe Slot with low-profile Form-Factor
- Powered by Hailo-8™ AI inference processors.
- Supports 104 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Introduction
SUNIX PCIe x16 AI Acceleration Card, AIEH4000, enables Hailo-8™ AI Processor into Edge computers, provides an easily deployable solution for AI developers to offload CPU loading on low-latency deep learning inference and cost-effective Edge AI computing with higher processing capabilities and lower power consumption. And this card can deliver up to 104 TOPS, enabling it to handle the most demanding computer vision applications. [Test Results: ResNet-50 v1( 224x224)@5,328FPS, YOLOv5m (640x640)@872FPS]
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
Features
SUNIX AIEH2000 VS NVIDIA A4000 AI visual computing acceleration performance comparison test(YOLOV5M)
SUNIX's AI accelerator card. This card not only significantly reduces computer energy consumption but also demonstrates exceptional performance in high-speed AI visual recognition computations. The AI accelerator card, AIEH2000, introduced by SUNIX, is a PCIe interface AI accelerator card specifically designed for PC. To showcase the performance and energy-saving capabilities of the SUNIX AI accelerator card, we conducted tests using the same testing platform as the Nvidia RTX A4000 graphics acceleration card, which already incorporates AI model inference capabilities. Employing the yolov5m model for image recognition tests allowed us to compare the performance of SUNIX's AIEH2000 with Nvidia A4000 and understand their differences in performance. The test results displayed on the screen vividly illustrate the data for both accelerator cards: SUNIX AIEH2000 achieves an average processing speed of approximately 400 frames per second (FPS), capable of handling up to 400 real-time frames per second. Meanwhile, the Nvidia A4000 processes approximately 300 frames per second (FPS), indicating its capability to handle around 300 frames per second. These test results indicate that utilizing SUNIX's AIEH2000 AI accelerator card for image recognition processing delivers a speed approximately 100 frames per second (FPS) faster than the Nvidia A4000 graphics card, equating to a computational speed increase of about 1.3 times. Regarding energy consumption, SUNIX's AIEH2000 demonstrates exceptional performance. During model inference computations, the AIEH2000 consumes an average of only 10 watts, while the Nvidia A4000 consumes approximately 60-100 watts. Based on testing and calculations, SUNIX Tech's AIEH2000 achieves a computation speed of 27 FPS per watt, compared to Nvidia A4000's mere 3 FPS per watt. Therefore, SUNIX's AI accelerator card not only excels in efficiency but also exhibits clear advantages in energy conservation and cost-effectiveness. Especially when considering the economic and environmental aspects of deploying AI edge computing frameworks on a large scale, the differences between the two become more pronounced.Specifications
Board Description |
|
| Model | AIEH4000 |
| Description | PCIe x16 AI Acceleration Card |
| AI Processors | Hailo-8™ AI Processor x4 |
| AI Performance | 104 TOPs (Tera-Operations Per Second) |
| Power Consumption | 27W (based on model) |
| PCIe Interface | PCI Express Gen3 x16 |
| AI Frameworks | TensorFLow, TensorFlow Lite, ONNX, Keras, Pytorch |
| Hailo Software Suites |
Dataflow Compiler (Model conversion and compilation) HailoRT (Runtime environment and driver) Model Zoo (Pre-trained models) TAPPAS (Deployment framework, examples and multi-network pipelines) |
| OS Support | Linux (e.g. Ubuntu, Yocto) & Windows 10 (X86/X64) / 11 |
| Certification | CE/FCC/VCCI/BSMI |
| Green Environment | RoHS |
Environment |
|
| Operation Temperature | 0 to 50°C (32 to 122°F) commercial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -20 to 85°C (-4 to 185°F) |
Dimension |
|
| PCB Dimension | 167.0 x 68.5 mm |
| Bracket | Standard 121 mm / Low profile 80mm(Optional) |
Support
Download Datasheet
| Datasheet | AIEH4000 Datasheet EN |
|
Download Guide
| Guide | AI Acceleration Card Quick Installation Guide |
|
Download Driver
| Driver | HailoRT - Windows | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:39:48 | |
| Size | 8.7MB | |
| OS |
Windows 10 64bit
Windows 11 64bit
|
| Driver | HailoRT - Linux - x86_64 | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:38:53 | |
| Size | 6.2MB | |
| OS |
Linux Kernel 5.x
|
| Driver | HailoRT PCIe Driver - Linux | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:37:31 | |
| Size | 132.8KB | |
| OS |
Linux Kernel 5.x
|
AIEH4000
PCIe x16 AI Accelerator Card
Hailo-8™ AI Processor x4
-
Compliant with PCI Express 3.0 x16
-
Supports Single PCIe Slot with low-profile Form-Factor
-
Powered by Hailo-8™ AI inference processors.
-
Supports 104 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Note: AIEH4000 is incompatible with AMD Platform when using x16 Link Width.
- Compliant with PCI Express 3.0 x16
- Supports Single PCIe Slot with low-profile Form-Factor
- Powered by Hailo-8™ AI inference processors.
- Supports 104 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Introduction
SUNIX PCIe x16 AI Acceleration Card, AIEH4000, enables Hailo-8™ AI Processor into Edge computers, provides an easily deployable solution for AI developers to offload CPU loading on low-latency deep learning inference and cost-effective Edge AI computing with higher processing capabilities and lower power consumption. And this card can deliver up to 104 TOPS, enabling it to handle the most demanding computer vision applications. [Test Results: ResNet-50 v1( 224x224)@5,328FPS, YOLOv5m (640x640)@872FPS]
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
Features
SUNIX AIEH2000 VS NVIDIA A4000 AI visual computing acceleration performance comparison test(YOLOV5M)
SUNIX's AI accelerator card. This card not only significantly reduces computer energy consumption but also demonstrates exceptional performance in high-speed AI visual recognition computations. The AI accelerator card, AIEH2000, introduced by SUNIX, is a PCIe interface AI accelerator card specifically designed for PC. To showcase the performance and energy-saving capabilities of the SUNIX AI accelerator card, we conducted tests using the same testing platform as the Nvidia RTX A4000 graphics acceleration card, which already incorporates AI model inference capabilities. Employing the yolov5m model for image recognition tests allowed us to compare the performance of SUNIX's AIEH2000 with Nvidia A4000 and understand their differences in performance. The test results displayed on the screen vividly illustrate the data for both accelerator cards: SUNIX AIEH2000 achieves an average processing speed of approximately 400 frames per second (FPS), capable of handling up to 400 real-time frames per second. Meanwhile, the Nvidia A4000 processes approximately 300 frames per second (FPS), indicating its capability to handle around 300 frames per second. These test results indicate that utilizing SUNIX's AIEH2000 AI accelerator card for image recognition processing delivers a speed approximately 100 frames per second (FPS) faster than the Nvidia A4000 graphics card, equating to a computational speed increase of about 1.3 times. Regarding energy consumption, SUNIX's AIEH2000 demonstrates exceptional performance. During model inference computations, the AIEH2000 consumes an average of only 10 watts, while the Nvidia A4000 consumes approximately 60-100 watts. Based on testing and calculations, SUNIX Tech's AIEH2000 achieves a computation speed of 27 FPS per watt, compared to Nvidia A4000's mere 3 FPS per watt. Therefore, SUNIX's AI accelerator card not only excels in efficiency but also exhibits clear advantages in energy conservation and cost-effectiveness. Especially when considering the economic and environmental aspects of deploying AI edge computing frameworks on a large scale, the differences between the two become more pronounced.Specifications
Board Description |
|
| Model | AIEH4000 |
| Description | PCIe x16 AI Acceleration Card |
| AI Processors | Hailo-8™ AI Processor x4 |
| AI Performance | 104 TOPs (Tera-Operations Per Second) |
| Power Consumption | 27W (based on model) |
| PCIe Interface | PCI Express Gen3 x16 |
| AI Frameworks | TensorFLow, TensorFlow Lite, ONNX, Keras, Pytorch |
| Hailo Software Suites |
Dataflow Compiler (Model conversion and compilation) HailoRT (Runtime environment and driver) Model Zoo (Pre-trained models) TAPPAS (Deployment framework, examples and multi-network pipelines) |
| OS Support | Linux (e.g. Ubuntu, Yocto) & Windows 10 (X86/X64) / 11 |
| Certification | CE/FCC/VCCI/BSMI |
| Green Environment | RoHS |
Environment |
|
| Operation Temperature | 0 to 50°C (32 to 122°F) commercial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -20 to 85°C (-4 to 185°F) |
Dimension |
|
| PCB Dimension | 167.0 x 68.5 mm |
| Bracket | Standard 121 mm / Low profile 80mm(Optional) |
Support
Download Datasheet
| Datasheet | AIEH4000 Datasheet EN |
|
Download Guide
| Guide | AI Acceleration Card Quick Installation Guide |
|
Download Driver
| Driver | HailoRT - Windows | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:39:48 | |
| Size | 8.7MB | |
| OS |
Windows 10 64bit
Windows 11 64bit
|
| Driver | HailoRT - Linux - x86_64 | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:38:53 | |
| Size | 6.2MB | |
| OS |
Linux Kernel 5.x
|
| Driver | HailoRT PCIe Driver - Linux | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:37:31 | |
| Size | 132.8KB | |
| OS |
Linux Kernel 5.x
|
AIEH4000
PCIe x16 AI Accelerator Card
Hailo-8™ AI Processor x4
-
Compliant with PCI Express 3.0 x16
-
Supports Single PCIe Slot with low-profile Form-Factor
-
Powered by Hailo-8™ AI inference processors.
-
Supports 104 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Note: AIEH4000 is incompatible with AMD Platform when using x16 Link Width.
- Compliant with PCI Express 3.0 x16
- Supports Single PCIe Slot with low-profile Form-Factor
- Powered by Hailo-8™ AI inference processors.
- Supports 104 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
SUNIX PCIe x16 AI Acceleration Card, AIEH4000, enables Hailo-8™ AI Processor into Edge computers, provides an easily deployable solution for AI developers to offload CPU loading on low-latency deep learning inference and cost-effective Edge AI computing with higher processing capabilities and lower power consumption. And this card can deliver up to 104 TOPS, enabling it to handle the most demanding computer vision applications. [Test Results: ResNet-50 v1( 224x224)@5,328FPS, YOLOv5m (640x640)@872FPS]
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
Introduction
SUNIX PCIe x16 AI Acceleration Card, AIEH4000, enables Hailo-8™ AI Processor into Edge computers, provides an easily deployable solution for AI developers to offload CPU loading on low-latency deep learning inference and cost-effective Edge AI computing with higher processing capabilities and lower power consumption. And this card can deliver up to 104 TOPS, enabling it to handle the most demanding computer vision applications. [Test Results: ResNet-50 v1( 224x224)@5,328FPS, YOLOv5m (640x640)@872FPS]
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
AIEH4000 also enables legacy computers such as NVRs, Edge AI boxes, Industrial PCs and robots to run video-intensive, mission-critical Edge AI applications such as video analytics, traffic management, access control.
Features
SUNIX AIEH2000 VS NVIDIA A4000 AI visual computing acceleration performance comparison test(YOLOV5M)
SUNIX's AI accelerator card. This card not only significantly reduces computer energy consumption but also demonstrates exceptional performance in high-speed AI visual recognition computations. The AI accelerator card, AIEH2000, introduced by SUNIX, is a PCIe interface AI accelerator card specifically designed for PC. To showcase the performance and energy-saving capabilities of the SUNIX AI accelerator card, we conducted tests using the same testing platform as the Nvidia RTX A4000 graphics acceleration card, which already incorporates AI model inference capabilities. Employing the yolov5m model for image recognition tests allowed us to compare the performance of SUNIX's AIEH2000 with Nvidia A4000 and understand their differences in performance. The test results displayed on the screen vividly illustrate the data for both accelerator cards: SUNIX AIEH2000 achieves an average processing speed of approximately 400 frames per second (FPS), capable of handling up to 400 real-time frames per second. Meanwhile, the Nvidia A4000 processes approximately 300 frames per second (FPS), indicating its capability to handle around 300 frames per second. These test results indicate that utilizing SUNIX's AIEH2000 AI accelerator card for image recognition processing delivers a speed approximately 100 frames per second (FPS) faster than the Nvidia A4000 graphics card, equating to a computational speed increase of about 1.3 times. Regarding energy consumption, SUNIX's AIEH2000 demonstrates exceptional performance. During model inference computations, the AIEH2000 consumes an average of only 10 watts, while the Nvidia A4000 consumes approximately 60-100 watts. Based on testing and calculations, SUNIX Tech's AIEH2000 achieves a computation speed of 27 FPS per watt, compared to Nvidia A4000's mere 3 FPS per watt. Therefore, SUNIX's AI accelerator card not only excels in efficiency but also exhibits clear advantages in energy conservation and cost-effectiveness. Especially when considering the economic and environmental aspects of deploying AI edge computing frameworks on a large scale, the differences between the two become more pronounced.Specifications
Board Description |
|
| Model | AIEH4000 |
| Description | PCIe x16 AI Acceleration Card |
| AI Processors | Hailo-8™ AI Processor x4 |
| AI Performance | 104 TOPs (Tera-Operations Per Second) |
| Power Consumption | 27W (based on model) |
| PCIe Interface | PCI Express Gen3 x16 |
| AI Frameworks | TensorFLow, TensorFlow Lite, ONNX, Keras, Pytorch |
| Hailo Software Suites |
Dataflow Compiler (Model conversion and compilation) HailoRT (Runtime environment and driver) Model Zoo (Pre-trained models) TAPPAS (Deployment framework, examples and multi-network pipelines) |
| OS Support | Linux (e.g. Ubuntu, Yocto) & Windows 10 (X86/X64) / 11 |
| Certification | CE/FCC/VCCI/BSMI |
| Green Environment | RoHS |
Environment |
|
| Operation Temperature | 0 to 50°C (32 to 122°F) commercial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -20 to 85°C (-4 to 185°F) |
Dimension |
|
| PCB Dimension | 167.0 x 68.5 mm |
| Bracket | Standard 121 mm / Low profile 80mm(Optional) |
Support
Download Datasheet
| Datasheet | AIEH4000 Datasheet EN |
|
Download Guide
| Guide | AI Acceleration Card Quick Installation Guide |
|
Download Driver
| Driver | HailoRT - Windows | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:39:48 | |
| Size | 8.7MB | |
| OS |
Windows 10 64bit
Windows 11 64bit
|
| Driver | HailoRT - Linux - x86_64 | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:38:53 | |
| Size | 6.2MB | |
| OS |
Linux Kernel 5.x
|
| Driver | HailoRT PCIe Driver - Linux | |
| Version | 4.18.0 | |
| Update | 2024-08-22 09:37:31 | |
| Size | 132.8KB | |
| OS |
Linux Kernel 5.x
|
