1 min read
This paper introduces a paradigm shift in storage architecture designed to overcome the CPU-centric data path bottlenecks in modern AI workloads. By combining NVIDIA SCADA (Scaled Accelerated Data Access) with a 100% liquid-cooled, fanless chassis housing 96 E3.S NVMe drives, we have created an...
1 min read
This paper explores a disaggregated key-value (KV) storage architecture designed to efficiently offload KV cache tensors for generative AI workloads. By synergizing Wiwynn's OCP ORv3-compliant servers with Pliops' hardware-accelerated data path, this framework delivers a highly scalable and...
1 min read
This paper explores an advanced framework designed to automate the extraction of important attributes from unstructured part datasheets. By synergizing expert data science preprocessing with the Retrieval-Augmented Generation (RAG) architecture, we have achieved industry-leading recognition...
Deploying large-scale AI clusters introduces engineering challenges that extend well beyond the individual server rack. From liquid cooling integration to high-voltage power distribution and network topology design, the "Last Mile" of AI infrastructure presents multifaceted integration complexities.
1 min read
This paper discusses the rapid expansion of AI workloads and the resulting transformation in data center infrastructure requirements. Traditional air-cooling systems are becoming less effective due to the increased power density of server racks. To address this challenge, the paper presents...
1 min read
Firmware updates are essential for the BMC system. Each device requires a unique update flow and utilizes different transport protocols, such as I2C or JTAG. In previous projects, various update utilities had to be developed to accommodate these requirements. However, without proper software...
1 min read
AI clusters using next-generation accelerators (e.g., NVIDIA GB200) push rack power density beyond 130 kW, making air cooling insufficient and driving adoption of direct liquid cooling (DLC). This whitepaper introduces Wiwynn’s Elastic Management Framework, a modular, scalable, interoperable...
1 min read
This whitepaper examines the growing importance of power efficiency in AI systems, where increasing computational demand translates into significant energy consumption and operating costs. We begin by introducing the overall architecture of AI applications and their critical power components,...
1 min read
The rapid expansion of artificial intelligence, high-performance computing, and cloud services is driving unprecedented levels of heat generation in modern data centers. Traditional air cooling and single-phase liquid cooling are increasingly unable to meet the demands of high-density server racks...