AI-MXM-S41A
산업용 보드/모듈 Sophgo BM1684 AI Processor
The AI-MXM-S41A is an MXM 3.1 Type A module equipped with the third generation TPU processor from Sophgo with 8GB of memory. The module provides up to 17.6 TOPS of AI performance at the typical power consumption of 30W.
Features
• Powered by Sophgo BM1684 AI Processor
• Supports Video Decoding of 1080P@960fps and H.265/H.264 Format
• Serves as the Small Form Factor Acceleration Platform of MXM Type A
• Drive Up to 17.6 TOPS AI Performance at Typical Power Consumption: 30W
• Supports Precision for INT8 and FP32
The AI-MXM-S41A is an MXM 3.1 Type A module equipped with the third generation TPU processor from Sophgo with 8GB of memory. The module provides up to 17.6 TOPS of AI performance at the typical power consumption of 30W. It also supports FP32 high precision and INT8 low precision, offering greater flexibility and performance optimization to meet various application demands. Higher computing efficiency and accuracy can also be achieved with smart management of computing resources in different scenarios. Additionally, INT8 models can be automatically compiled and calibrated, providing comprehensive toolchain support for deep learning frameworks such as Caffe, Mxnet, TensorFlow, Pytorch, and Paddle.
Due to the reduction in size, weight, and power consumption, AI-MXM-S41A can be easily integrated into various embedded systems for low-latency processing of heavy inference workloads. Users can shorten the development cycle of AI projects through comprehensive software packages, such as AI development tools, industrial-grade mechanism design as well as the fine-tuning of AI inference integration model and further customized services.