Apple announced that multiple new Mac Studio units with the M5 Ultra chip can be clustered over Thunderbolt 5 and RDMA to create a shared memory pool for running larger AI models. Clustering four syst…
Apple's MLX team introduced distributed inference and training capabilities for machine learning workloads across multiple Macs at WWDC26, demonstrating how to scale large language models using RDMA o…
NVIDIA's NVLink and NVSwitch technologies form a scale-up fabric that connects GPUs tightly enough to behave as a single machine, contrasting with scale-out fabrics like InfiniBand and RoCE that link …
Scality is enhancing its object storage to support AI workloads by integrating with NVIDIA's NIXL library, enabling S3 over RDMA to bypass CPU bottlenecks and move data directly between flash and GPU …
Antirez achieved running GLM 5.2 and DeepSeek v4 Flash with Tensor Parallelism across two M5Max 128GB MacBooks via RDMA, enabling models that previously could not fit on any affordable machine to run …
NANOG 97 in Bellevue, Washington, focused on AI's impact on data center design and network operations, with Cogent's Dave Schaeffer presenting statistics showing US AI infrastructure spending will rea…
AI clusters often underperform despite powerful GPUs because the GPUs are idle due to bottlenecks in data loading, CPU preprocessing, network communication, or storage contention. A developer explains…
An AI agent tasked with optimizing a system's performance achieved full NIC and disk bandwidth saturation, but its explanation of the optimization was completely wrong. The system, which uses a single…
Microsoft announced a broad set of enhancements to Azure Kubernetes Service at Build 2026, including bare metal support, fleet management, and AI infrastructure features, aiming to make Kubernetes a f…
VDURA has enhanced its PanFS parallel file system with a new control plane featuring multi-tenancy, faster S3 object storage performance, and native S3 object tagging to better support enterprise and …
InfiniBand, a high-performance interconnect technology designed for Remote Direct Memory Access (RDMA), has become critical for AI training and inference workloads that require direct data movement be…
Canonical published an explainer on RDMA over Converged Ethernet (RoCE), a technology that runs the RDMA programming model over standard Ethernet networks to bypass the kernel and reduce CPU involveme…
Intel launched the Ethernet E835 Controllers and Network Adapters, offering up to 200GbE bandwidth for cloud, edge, and enterprise data centers. The new 800 Series products feature PCIe 5.0, RDMA, and…
Researchers from UC Berkeley's UCCL project released mKernel, a library of persistent CUDA kernels that fuse intra-node NVLink communication, inter-node RDMA, and compute into a single kernel to addre…
Hugging Face researchers released a new method called delta weight sync that reduces the per-step weight transfer in asynchronous reinforcement learning from 1.2 GB to as little as 20 MB for a 0.6B pa…