Scott Baker, IBM | Nvidia GTC 2024
summary
IBM says Storage Scale delivers 4.5 PB in 4U as a distributed file/object system and claims 70% power reduction versus prior generations, with entry 3500 and larger 6000 SKUs for rapid on‑prem AI. Watsonx.ai/.data/.governance runs containerized on Red Hat OpenShift atop IBM Fusion as a turnkey stack, integrated with NVIDIA GPUDirect and GPU‑as‑a‑service. FlashCore Module 4 pushes ML onto drives for hardware‑accelerated inline threat detection and faster automated responses.
AI-written summary · about
chapters (7)
- 0:00AI infrastructure impact
- 1:20Architectural storage shift
- 3:24Software defined storage
- 5:44IBM and Nvidia partnership
- 8:46Future of AI storage
- 11:28Storage solutions for AI
- 13:36Computational storage
related talks
17mJeff Denworth, VAST Data | Databricks Data + AI Summit 2023
48mScale and accelerate the impact of generative AI with watsonx
47mBeyond Agents: What AI Strategy Really Needs in 2025
28mScaling agentic AI: From infrastructure to enterprise outcomes
05mFrom Transactions to Transformers: Hardware Built for...
10mApproach to AI at a large scale
23mScaling Up Your Cloud with Oracle Superclusters
28mShared Yet Isolated at Scale: Building Multi-Tenant Inference Platform on…