gpu-clusters#
54 picked talks tagged gpu-clusters, newest conference first.
NANOG 98#
PyTorch Conference 2026#
- Introducing NCCL Extensions: Communication Patterns for Modern AI
- MCCL: Fault-Tolerant Collective Communication for Large-Scale GPU Training
- rocSHMEM Symmetric Memory in PyTorch for AMD GPUs
- State-of-the-Art KV Transfer for Disaggregated LLM Serving in vLLM
- XCCL: Scaling PyTorch Collectives to Exascale on Intel GPUs with TorchComms
OCP Global Summit 2026#
- ESUN: Open Ethernet for the AI Scale-Up Era
- SUE-T - An Ethernet transport for Scale-Up Networking
- Tahan: An Open, Ethernet-Based Scale-Up Switch for Meta's MTIA Accelerators
OPI Summit on DPUs/IPUs 2026#
AI Engineer New York 2026#
- Compute Exchange
- How Compute Markets Work
- Not All GPU's Are Deployed The Same
- Shipping like a startup, regulated like a bank
- The Dark Pools of Compute
- Your GPU Neighbor Can Hear Your Model
@Scale: Networking 2026#
IETF 126 (Vienna)#
- Fully Adaptive Routing Ethernet in Multi-Plane Scale-Out Networks
- SRv6 for Deterministic Path Placement in AI Backends
AI Engineer World's Fair 2026#
- Can LLMs write fast multi-GPU kernels? We built a benchmark to find out.
- Fault-Tolerant Training at Scale: Making Hardware Failures a Non-Event
- Modular: Taming the AI Hardware Cambrian Explosion
@Scale: Systems & Reliability 2026#
- AI Storage Blueprint
- Experience Productionizing and Operating GB200 clusters
- Extending Meta to the Public Cloud
- Taming AI Infrastructure Failures with Agentic Debugging
@Scale: AI & Data 2026#
nog.fi meeting 2026.06#
CHNUG #3#
MLSys 2026#
- TokenWeave: Efficient Compute-Communication Overlap for Distributed LLM Inference
- Unleashing Scalable Context Parallelism for Foundation Models Pre-Training via FCP
KubeCon + CloudNativeCon Europe 2026#
- KubeVirt on GB200: Virtualizing a Rack-Scale Supercomputer
- Making Topology-Aware Scheduling Practical for AI Workloads: From Discovery to Simulation at Scale
- The Latest in GPU, TPU, NIC and Other Device Support — WG Device Management
- Virtualizing Large Scale GPU Cluster for Sovereign AI: Petasus AI Cloud Journey with Kubernetes
IETF 125 (Shenzhen)#
GTC 2026#
- Achieve Truly Serverless GPUs With libfuse, CRIU, and CUDA-Checkpoint
- Break Communication Barriers: Scale AI and HPC with NCCL, NIXL, and NVSHMEM
- Build a Software-Defined Multi-Tenant NVLinked Cluster
- Building Sovereign AI: Scaling 100B+ Model Training on NVIDIA Blackwell Infrastructure (Presented by Lablup, Inc.)
- Continuous Inference Benchmarking of the AI Frontier
- Enhance Data Center Efficiency and Throughput by Co-Scheduling on NVIDIA CPU+GPU Systems
- From Kernels to Clusters: How PyTorch Powers High-Performance AI (Presented by Meta)
- Solve the Day 1 Challenge of Landing LLM Inference in Enterprise
Cisco Live Europe 2026 (Amsterdam)#
OVS + OVN 2025 Fall Conference (Prague)#
KubeCon + CloudNativeCon North America 2025#
OCP Global Summit 2025#
MLSys 2025#
- COMET: Fine-grained Computation-communication Overlapping for Mixture-of-Experts
- Context Parallelism for Scalable Million-Token Inference
- TileLink: Generating Efficient Compute-Communication Overlapping Kernels using Tile-Centric Primitives