AIMH1000 (OEM only)
M.2 M key AI Acceleration module
Hailo-8™ AI Processor x1
-
Compliant with PCI Express 3.0 x4
-
Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
-
Powered by Hailo-8™ AI inference processors.
-
Supports 26 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
- Compliant with PCI Express 3.0 x4
- Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
- Powered by Hailo-8™ AI inference processors.
- Supports 26 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
.png)
Introduction
The AIMH1000 AI acceleration module, compatible with M.2 Key M form factor, is a 26 TOPS acceleration module, supports Edge AI applications in computer vision. [Test Results: ResNet-50 v1( 224x224)@1,332FPS, YOLOv5m (640x640)@218FPS]
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
Features
SUNIX AIEH1000 V.S Intel i7-12900 Performance comparison
Today, I'd like to show you how to significantly boost the speed of computer image recognition on a PC by using the SUNIX AI Accelerator Card. Next, I will demonstrate our latest product, the AIEH1000, which is a dedicated PCIe interface AI accelerator card developed for PCs. The testing platform is the X86, equipped with an Intel i7-12900 processor. While this processor already come with AI inference acceleration capabilities, we will conduct a comparative test using SUNIX Technology's AIEH1000 AI accelerator card on the same computer to clearly showcase the performance difference. Let’s , take a look at the screen. For our image recognition tests, we are using the pre-trained YOLOv5m model. First, let's examine the performance of the CPU processor. The screen on the left displays the CPU performance results, which show that it can only process 7 frames per second (FPS), meaning it can handle just 7 frames per second. Next, let's see the results when we insert the SUNIX AIEH1000 AI Accelerator Card. Now, the test results are shown on the right side of the screen, with an average speed of 218 frames per second (FPS), which means it’s capable of processing up to 218 real-time frames per second. From the test results, it's clear that the image processing speed with the SUNIX AIEH1000 AI accelerator card is more than 30 times faster than the CPU processor. So, in the context of computer vision applications and the rapidly growing field of AI, a standard PC equipped with the SUNIX AI accelerator card can better meet the expectations of customers and the market. Moreover, our AI accelerator card can reduce the CPU load on your system, satisfying customer demands for real-time streaming of image processing and multitasking model processing. As to energy efficiency , our AIEH1000 AI accelerator card only consumes 5W in average during model inference, making it suitable for implementing edge AI applications on your PC. You can now readily adopt the SUNIX AI Accelerator Card to expand your PC business into the AI visual application marketSpecifications
Board Description |
|
| Model | AIMH1000 |
| Description | M.2 M key AI Acceleration module |
| AI Processors | Hailo-8™ AI Processor x1 |
| AI Performance | 26 TOPs (Tera-Operations Per Second) |
| Power Consumption | 8W (depend on heatsink design) |
| PCIe Interface | PCI Express Gen3 x4 |
| 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 Class A |
Environment |
|
| Operation Temperature | -45 to 85°C (-49 to 185°F) industrial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -45 to 85°C (-49 to 185°F) |
Dimension |
|
| PCB Dimension | 22x42 / 22x60 / 22x80 mm |
| Form Factor | M.2 Key M |
| Note: Proper heat dissipation must be employed to ensure that the Hailo-8 chip does not overheat. The module requests a heatsink based on platform thermal case. | |
Support
Download Datasheet
| Datasheet | AIMH1000 Datasheet EN |
|
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
|
AIMH1000 (OEM only)
M.2 M key AI Acceleration module
Hailo-8™ AI Processor x1
-
Compliant with PCI Express 3.0 x4
-
Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
-
Powered by Hailo-8™ AI inference processors.
-
Supports 26 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
- Compliant with PCI Express 3.0 x4
- Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
- Powered by Hailo-8™ AI inference processors.
- Supports 26 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Introduction
The AIMH1000 AI acceleration module, compatible with M.2 Key M form factor, is a 26 TOPS acceleration module, supports Edge AI applications in computer vision. [Test Results: ResNet-50 v1( 224x224)@1,332FPS, YOLOv5m (640x640)@218FPS]
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
Features
SUNIX AIEH1000 V.S Intel i7-12900 Performance comparison
Today, I'd like to show you how to significantly boost the speed of computer image recognition on a PC by using the SUNIX AI Accelerator Card. Next, I will demonstrate our latest product, the AIEH1000, which is a dedicated PCIe interface AI accelerator card developed for PCs. The testing platform is the X86, equipped with an Intel i7-12900 processor. While this processor already come with AI inference acceleration capabilities, we will conduct a comparative test using SUNIX Technology's AIEH1000 AI accelerator card on the same computer to clearly showcase the performance difference. Let’s , take a look at the screen. For our image recognition tests, we are using the pre-trained YOLOv5m model. First, let's examine the performance of the CPU processor. The screen on the left displays the CPU performance results, which show that it can only process 7 frames per second (FPS), meaning it can handle just 7 frames per second. Next, let's see the results when we insert the SUNIX AIEH1000 AI Accelerator Card. Now, the test results are shown on the right side of the screen, with an average speed of 218 frames per second (FPS), which means it’s capable of processing up to 218 real-time frames per second. From the test results, it's clear that the image processing speed with the SUNIX AIEH1000 AI accelerator card is more than 30 times faster than the CPU processor. So, in the context of computer vision applications and the rapidly growing field of AI, a standard PC equipped with the SUNIX AI accelerator card can better meet the expectations of customers and the market. Moreover, our AI accelerator card can reduce the CPU load on your system, satisfying customer demands for real-time streaming of image processing and multitasking model processing. As to energy efficiency , our AIEH1000 AI accelerator card only consumes 5W in average during model inference, making it suitable for implementing edge AI applications on your PC. You can now readily adopt the SUNIX AI Accelerator Card to expand your PC business into the AI visual application marketSpecifications
Board Description |
|
| Model | AIMH1000 |
| Description | M.2 M key AI Acceleration module |
| AI Processors | Hailo-8™ AI Processor x1 |
| AI Performance | 26 TOPs (Tera-Operations Per Second) |
| Power Consumption | 8W (depend on heatsink design) |
| PCIe Interface | PCI Express Gen3 x4 |
| 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 Class A |
Environment |
|
| Operation Temperature | -45 to 85°C (-49 to 185°F) industrial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -45 to 85°C (-49 to 185°F) |
Dimension |
|
| PCB Dimension | 22x42 / 22x60 / 22x80 mm |
| Form Factor | M.2 Key M |
| Note: Proper heat dissipation must be employed to ensure that the Hailo-8 chip does not overheat. The module requests a heatsink based on platform thermal case. | |
Support
Download Datasheet
| Datasheet | AIMH1000 Datasheet EN |
|
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
|
AIMH1000 (OEM only)
M.2 M key AI Acceleration module
Hailo-8™ AI Processor x1
-
Compliant with PCI Express 3.0 x4
-
Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
-
Powered by Hailo-8™ AI inference processors.
-
Supports 26 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
- Compliant with PCI Express 3.0 x4
- Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
- Powered by Hailo-8™ AI inference processors.
- Supports 26 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
Introduction
The AIMH1000 AI acceleration module, compatible with M.2 Key M form factor, is a 26 TOPS acceleration module, supports Edge AI applications in computer vision. [Test Results: ResNet-50 v1( 224x224)@1,332FPS, YOLOv5m (640x640)@218FPS]
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
Features
SUNIX AIEH1000 V.S Intel i7-12900 Performance comparison
Today, I'd like to show you how to significantly boost the speed of computer image recognition on a PC by using the SUNIX AI Accelerator Card. Next, I will demonstrate our latest product, the AIEH1000, which is a dedicated PCIe interface AI accelerator card developed for PCs. The testing platform is the X86, equipped with an Intel i7-12900 processor. While this processor already come with AI inference acceleration capabilities, we will conduct a comparative test using SUNIX Technology's AIEH1000 AI accelerator card on the same computer to clearly showcase the performance difference. Let’s , take a look at the screen. For our image recognition tests, we are using the pre-trained YOLOv5m model. First, let's examine the performance of the CPU processor. The screen on the left displays the CPU performance results, which show that it can only process 7 frames per second (FPS), meaning it can handle just 7 frames per second. Next, let's see the results when we insert the SUNIX AIEH1000 AI Accelerator Card. Now, the test results are shown on the right side of the screen, with an average speed of 218 frames per second (FPS), which means it’s capable of processing up to 218 real-time frames per second. From the test results, it's clear that the image processing speed with the SUNIX AIEH1000 AI accelerator card is more than 30 times faster than the CPU processor. So, in the context of computer vision applications and the rapidly growing field of AI, a standard PC equipped with the SUNIX AI accelerator card can better meet the expectations of customers and the market. Moreover, our AI accelerator card can reduce the CPU load on your system, satisfying customer demands for real-time streaming of image processing and multitasking model processing. As to energy efficiency , our AIEH1000 AI accelerator card only consumes 5W in average during model inference, making it suitable for implementing edge AI applications on your PC. You can now readily adopt the SUNIX AI Accelerator Card to expand your PC business into the AI visual application marketSpecifications
Board Description |
|
| Model | AIMH1000 |
| Description | M.2 M key AI Acceleration module |
| AI Processors | Hailo-8™ AI Processor x1 |
| AI Performance | 26 TOPs (Tera-Operations Per Second) |
| Power Consumption | 8W (depend on heatsink design) |
| PCIe Interface | PCI Express Gen3 x4 |
| 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 Class A |
Environment |
|
| Operation Temperature | -45 to 85°C (-49 to 185°F) industrial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -45 to 85°C (-49 to 185°F) |
Dimension |
|
| PCB Dimension | 22x42 / 22x60 / 22x80 mm |
| Form Factor | M.2 Key M |
| Note: Proper heat dissipation must be employed to ensure that the Hailo-8 chip does not overheat. The module requests a heatsink based on platform thermal case. | |
Support
Download Datasheet
| Datasheet | AIMH1000 Datasheet EN |
|
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
|
AIMH1000 (OEM only)
M.2 M key AI Acceleration module
Hailo-8™ AI Processor x1
-
Compliant with PCI Express 3.0 x4
-
Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
-
Powered by Hailo-8™ AI inference processors.
-
Supports 26 TOPS AI performance.
-
Low power consumption.
-
Supports Software Development Kit.
-
Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
- Compliant with PCI Express 3.0 x4
- Supports M.2 Key M Form-Factor with 2242, 2260 and 2280 adjustable
- Powered by Hailo-8™ AI inference processors.
- Supports 26 TOPS AI performance.
- Low power consumption.
- Supports Software Development Kit.
- Real-time, Low latency and High-performance AI inference acceleration to Edge AI.
The AIMH1000 AI acceleration module, compatible with M.2 Key M form factor, is a 26 TOPS acceleration module, supports Edge AI applications in computer vision. [Test Results: ResNet-50 v1( 224x224)@1,332FPS, YOLOv5m (640x640)@218FPS]
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
Introduction
The AIMH1000 AI acceleration module, compatible with M.2 Key M form factor, is a 26 TOPS acceleration module, supports Edge AI applications in computer vision. [Test Results: ResNet-50 v1( 224x224)@1,332FPS, YOLOv5m (640x640)@218FPS]
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
The module is based on PCIe gen3 x4 lanes interface which enables high throughput of input and output data, and it has adjustable form-factors of 2242, 2260 and 2280.
As high performance PCIe device, it can be used to perform real-time and low latency neural network inference. It uses PCIe interface for streaming input data and also for streaming inference results.
Features
SUNIX AIEH1000 V.S Intel i7-12900 Performance comparison
Today, I'd like to show you how to significantly boost the speed of computer image recognition on a PC by using the SUNIX AI Accelerator Card. Next, I will demonstrate our latest product, the AIEH1000, which is a dedicated PCIe interface AI accelerator card developed for PCs. The testing platform is the X86, equipped with an Intel i7-12900 processor. While this processor already come with AI inference acceleration capabilities, we will conduct a comparative test using SUNIX Technology's AIEH1000 AI accelerator card on the same computer to clearly showcase the performance difference. Let’s , take a look at the screen. For our image recognition tests, we are using the pre-trained YOLOv5m model. First, let's examine the performance of the CPU processor. The screen on the left displays the CPU performance results, which show that it can only process 7 frames per second (FPS), meaning it can handle just 7 frames per second. Next, let's see the results when we insert the SUNIX AIEH1000 AI Accelerator Card. Now, the test results are shown on the right side of the screen, with an average speed of 218 frames per second (FPS), which means it’s capable of processing up to 218 real-time frames per second. From the test results, it's clear that the image processing speed with the SUNIX AIEH1000 AI accelerator card is more than 30 times faster than the CPU processor. So, in the context of computer vision applications and the rapidly growing field of AI, a standard PC equipped with the SUNIX AI accelerator card can better meet the expectations of customers and the market. Moreover, our AI accelerator card can reduce the CPU load on your system, satisfying customer demands for real-time streaming of image processing and multitasking model processing. As to energy efficiency , our AIEH1000 AI accelerator card only consumes 5W in average during model inference, making it suitable for implementing edge AI applications on your PC. You can now readily adopt the SUNIX AI Accelerator Card to expand your PC business into the AI visual application marketSpecifications
Board Description |
|
| Model | AIMH1000 |
| Description | M.2 M key AI Acceleration module |
| AI Processors | Hailo-8™ AI Processor x1 |
| AI Performance | 26 TOPs (Tera-Operations Per Second) |
| Power Consumption | 8W (depend on heatsink design) |
| PCIe Interface | PCI Express Gen3 x4 |
| 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 Class A |
Environment |
|
| Operation Temperature | -45 to 85°C (-49 to 185°F) industrial |
| Operation Humidity | 5 to 95% RH |
| Storage Temperature | -45 to 85°C (-49 to 185°F) |
Dimension |
|
| PCB Dimension | 22x42 / 22x60 / 22x80 mm |
| Form Factor | M.2 Key M |
| Note: Proper heat dissipation must be employed to ensure that the Hailo-8 chip does not overheat. The module requests a heatsink based on platform thermal case. | |
Support
Download Datasheet
| Datasheet | AIMH1000 Datasheet EN |
|
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
|
