175: GreyBeards Talk Accelerated Object with SNIA TWG CoChairs, Jason Goldschmidt, DELL Distinguished Eng. & Nick Connolly, ARM Principal Eng.
In this episode, Dell Distinguished Engineer Jason Goldschmidt and ARM Principal Engineer Nick Connolly discuss the SNEA Accelerated Object Working Group’s effort to bring RDMA‑accelerated, S3‑compatible object storage to AI workloads. They explain how traditional object storage, while scalable, struggles with the latency and throughput demands of training and especially inference, where GPU memory limits and KV‑cache offloading become critical. By leveraging zero‑copy RDMA, they aim to reduce CPU and GPU overhead, cut latency, and enable high‑throughput data pipelines for both training data staging and inference serving. Their perspective combines deep industry experience in storage protocols with a forward‑looking view of AI’s exploding data needs.
174: GreyBeards Talk SDN Chips with Ted Weatherford, VP Bus. Dev. & John Carney. Dist. Eng. At Xsight Labs
In this episode, Ted Weatherford and John Carney of Xsight Labs discuss their SDN‑focused ASICs—the 12.8 Tbps X2 switch and the 800 Gbps E1 DPU—highlighting their low‑power, highly programmable designs built on TSMC N5. They explain how the X2’s 3072 MAP cores and...