2020.10.15

视频分析 - 精细调整以获得重大收益

分享：

ALL_news_tech_blog_26A07_xpXErIijiW

视频分析 – 精细调整以获得重大收益

AEWIN一直努力让您的硬体投资发挥最高效能。我们的任务是提供GPU加速的MEC 边缘伺服器针对视频分析应用。特别是加速的实时交通监控，利用NVIDIAT4解码实时视频流并进行人工智能推断，以检测不同类型车辆的交通模式、碰撞、交通违规、车牌和其他有用的指标。

这似乎是一个简单的任务，伺服器可以轻松处理。然而，并非所有伺服器都是一样的。现成的伺服器通常针对各种不同的工作负载进行优化，使其成为万能工具却无法精通任何一项。我们的 MEC 伺服器和工程师一直专注于最大化 GPU 加速性能。我们因能够处理而获得客户的赞誉。24x 全高清直播串流与他们之前的系统相比，该系统仅能处理18个流，使用类似的设置。通过持续的测试和微调，我们已经超越了一些伺服器领域中最大的名字的性能，并与NVIDIA发布的基准数据非常接近。

ResNet推理测试

表格：Resnet-50v1.5 + TensorRT 推断测试
*测试设置：SCB-1921 MEC 版本 + 2x Nvidia T4 与 Nvidia 和竞争对手的已发佈数据

这是选择其他制造商的一般用途伺服器与选择AEWIN之间的区别。通过个性化的合作，我们能够为您的特定工作负载微调伺服器和GPU的性能。我们的灵活制造位于新北市的工程/总部旁边，使我们能够灵活地与客户合作，从小型项目到大型项目。我们经验丰富的工程师随时可用，并能提供见解和系统配置建议。在决定您的下一个GPU加速平台之前，务必将您的项目带到AEWIN，看看AEWIN如何帮助您最大化投资的性能。

SCB-1921
边缘计算伺服器

支援双第二代英特尔至强可扩展处理器
支持2933MHz系统记忆体
4个PCIe x8用于网络扩展模块，另外4个PCIe x8或2个PCIe x16用于GPU或FPGA

Rack-Scale AI Infrastructure: Maximizing Performance, Efficiency, and Scalability for the AI Era

Driven by the explosion of Gen AI, Agentic AI, and the massive datasets behind them, computing infrastructure is evolving from standalone servers to rack-scale architectures. Modern AI workloads require a tightly integrated combination of computing, networking, storage, and cooling solutions to deliver maximum performance and efficiency. Future-Ready AI Infrastructure has become the foundation for the AI Era.

2026.06.30

Enhancing Network Resilience with AEWIN Gen4 LAN Bypass

Traditional LAN bypass focuses on keeping traffic flowing when a system goes down, but modern deployments require greater flexibility to balance availability and security. AEWIN Gen4 LAN bypass builds on the Gen3 foundation by introducing enhanced traffic control mechanisms to enable network behavior to better align with real-world operational demands.

2026.06.30

Optimizing Thermal Design for High-Performance Network Appliances and Servers

As modern data centers and network infrastructures continue to scale, the demand for higher computing performance is rapidly increasing. This trend drives CPU power consumption to new levels, especially with the latest server-grade processors. As a result, optimized thermal management has become a critical design factor that directly impacts system stability and performance. High-performance network appliances and servers require advanced cooling solutions to sustain performance under heavy workloads.

视频分析 - 精细调整以获得重大收益

相关讯息

Rack-Scale AI Infrastructure: Maximizing Performance, Efficiency, and Scalability for the AI Era

Enhancing Network Resilience with AEWIN Gen4 LAN Bypass

Optimizing Thermal Design for High-Performance Network Appliances and Servers