SOSP 2026

Meta Info

Homepage: https://sigops.org/s/conferences/sosp/2026/

Paper list: https://sigops.org/s/conferences/sosp/2026/accepted.html

Acceptance Rate

15.9% (= 62 / 390)

Papers

Large Language Models (LLMs)

• LLM Training

  • A Few GPUs, A Whole Lotta Scale: Faithful LLM Training Emulation with CrystalLLM

    • ZJU & Harvard & Alibaba

• LLM Inference

  • Offline Inference

    • Batched in Back: Characterizing and Optimizing Offline LLM Inference in Production with ACDC

      • SJTU & Alibaba

  • Multi-LLM Serving

    • Multi-LLM Serving at Production Scale [Code]

      • PKU & JD & UCAS

  • Wafer-Scale Inference

    • MeshRT: Compile-Time Governed Wafer-Scale Runtime for Low-Latency High-Throughput Inference

      • Edinburgh & MSRA

    • Wavel: A Fast and Efficient Compilation System for Wafer-Scale Accelerators

      • MSRA

• LLM Applications

  • Agentic Systems

    • Skill VM: Write Once, Run Everywhere Efficiently [arXiv] [Code] [Homepage]

      • SJTU IPADS

      • Introduce SkVM, treating agent skills as programs and LLMs plus harnesses as heterogeneous processors to improve skill portability across models and agent runtimes.

      • Combine AOT compilation based on model-harness capability profiling with runtime JIT optimization, including code solidification and feedback-driven skill rewriting.

      • Evaluate across eight LLMs and three agent harnesses; improve task completion by about 10%, reduce token cost by up to 40%, and cut end-to-end latency by 19-50x through JIT-boost code solidification.

Deep Learning Compilation

• GPU Kernels and Compilers

  • Taming Dynamism on GPUs: Cross-SM Kernel Fusion via SM Cooperation and Just-in-Time Reduction

    • SJTU IPADS

    • Introduce MorphKernel, targeting dynamic and imbalanced GPU operators such as decoding attention and MoE expert activation.

    • Fuse multi-phase kernels across SMs by decomposing coarse phase barriers into sub-task dependencies with fine-grained SM synchronization and on-chip data exchange, reducing HBM traffic for intermediate results.

    • Provide a declarative programming abstraction over operators such as map, iter, and reduce, while the runtime dynamically decides scheduling, communication, and parallel partitioning from input shapes.

    • Use TMA-based send/recv primitives, GPC-topology-aware communication, and per-SM instruction queues; improve performance by 1.3x on average over real workloads.

Diffusion Models

• Diffusion Model Serving

  • DiFlow: A System for Micro-Serving Text-to-Image Diffusion Workflows [arXiv]

    • HKUST & Alibaba

    • Our work!

    • Decomposes text-to-image diffusion workflows into independently managed model-execution nodes.

    • Enables per-model scaling, cross-workflow model sharing, and adaptive model parallelism for bursty serving workloads.