[SOSP 2025: The 31st Symposium on Operating Systems Principles](https://sigops.org/s/conferences/sosp/2025/accepted.html) - **Device-Assisted Live Migration of RDMA Devices** - **PhoenixOS: Concurrent OS-level GPU Checkpoint and Restore with Validated Speculation** - **Pie: A Programmable Serving System for Emerging LLM Applications** - **Aegaeon: Effective GPU Pooling for Concurrent LLM Serving on the Market** - **LithOS: An Operating System for Efficient Machine Learning on GPUs** - **cache_ext: Customizing the Page Cache with eBPF** - **AutoMan: Facilitating Verified Distributed Systems Development Through Automatic Code Generation and Manual Optimizations** - **Jenga: Effective Memory Management for Serving LLM with Heterogeneity** - **C-Cache: Efficient Large Language Model Serving via In-context Caching** - **PrefillOnly: An Inference Engine for Prefill-only Workloads in Large Language Model Applications** - **Robust LLM Training Infrastructure at ByteDance** - **KTransformers: Unleashing the Full Potential of CPU/GPU Hybrid Inference for MoE Models** - **Mycroft: Tracing Dependencies in Collective Communication Towards Reliable LLM Training** - **Loom: Efficient Capture and Querying of High-Frequency Telemetry** - **Managing Scalable Direct Storage Accesses for GPUs with GoFS**