cs.DCNov 5, 2026

KOSMOS: Self-scheduling Persistent Kernels for Communication-Compute Overlap

Authors: Alexander Magro, Aristotle Martin, Ye Wang, Wen Chen

Organizations: Advanced Micro Devices Inc., Austin, TX, USA.

Abstract

Tensor and sequence parallelism (TP+SP) and expert parallelism (EP) put collectives on the critical path of transformer layers. State-of-the-art fused kernels mitigate this with fixed roles or schedules, but this rigidity leaves large idle windows. Predicting a communication-computation split does not remove them either; a predetermined split cannot follow data readiness. We present KOSMOS, a library for self-scheduling persistent kernels whose workgroups use queues ordered by consumer demand to switch between communication and computation roles as data become ready, with no scheduler, per-call cost model, or tuned split. KOSMOS fuses every TP+SP collective of a transformer layer with its GEMM, and runs an entire expert layer, from router to fold, in one launch per direction. Evaluated on AMD Instinct MI350X and AMD Instinct MI355X GPUs, KOSMOS speeds up TP+SP end-to-end training in Megatron-LM by 1.10-1.28x in BF16 and 1.10-1.25x in MXFP8. KOSMOS accelerates end-to-end training of DeepSeek-V3 and Qwen3-235B proxies by 1.29-1.83x over DeepEP.

Figures & tables

Appendix figures & tables30 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. mKernel: Fast Multi-GPU, Multi-Node Fused Kernels

    Sep 11, 2026Ziming Mao, Yihan Zhang, Shawn Wei Chew +5GPU Kernel OptimizationCommunication-Efficient Distributed Training

  2. CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs

    May 19, 2026Han Guo, Jack Zhang, Arjun Menon +4GPU Kernel OptimizationGPU Acceleration