WEKA的新NeurMesh利用NVIDIA的Blue Field-4 DPU来操作人工智能和储存工作量更快、更便宜、更有效率。 WEKA’s new NeuralMesh uses NVIDIA’s BlueField-4 DPU to run AI and storage workloads faster, cheaper, and more efficiently.

WEKA为NVIDIA的Blue Field Field-4 DPU推出了新的NeurMesh结构,使储存和AI工作量能够直接在DPU而不是单独的CPU上运行。 WEKA has launched a new NeuralMesh architecture optimized for NVIDIA’s BlueField-4 DPU, enabling storage and AI workloads to run directly on the DPU instead of separate CPUs. 集成使用Blue Field-4的800 Gb/s网络,并增加计算,以提高性能,削减电力使用——每瓦增加100多个标志,用于人工智能代理,并降低基础设施成本。 The integration uses BlueField-4’s 800 Gb/s networking and enhanced compute to boost performance, cut power use—over 100x more tokens per watt for agentic AI—and reduce infrastructure costs. 该系统提供可扩展、安全和精简的AI业务,与荷兰国际数据研究所关于统一数据中心基础设施的愿景保持一致。 The system offers scalable, secure, and streamlined AI operations, aligning with NVIDIA’s vision for unified data center infrastructure.