mlsanityData to 7 Oct 2026

Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel

arXiv 2604.13327 ↗First posted 14 Apr 2026cs.DC cs.LG cs.PLcited by 4

Not on the map. Why?

Abstract (arXiv)

Modern GPU workloads, especially large language model (LLM) inference, suffer from kernel launch overheads and coarse synchronization that limit inter-kernel parallelism. Recent megakernel techniques fuse multiple operators into a single persistent kernel to eliminate launch gaps and expose inter-kernel parallelism, but struggle to handle dynamic shapes and data-dependent computation in real workloads. We present Event Tensor, a unified compiler abstraction for dynamic megakernels. Event Tensor encodes dependencies between tiled tasks, and enables first-class support for both shape and data-dependent dynamism. Built atop this abstraction, our Event Tensor Compiler (ETC) applies static and dynamic scheduling transformations to generate high-performance persistent kernels. Evaluations show that ETC achieves state-of-the-art LLM serving latency while significantly reducing system warmup overhead.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 21 of its 42 references that are in the map. First posted on arXiv 14 Apr 2026.

Builds on

21 of 42 references.

References · 21

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  3. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
  4. 2019Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelismcited by 1,829
  5. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,107
  6. 2024DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Servingcited by 595
  7. 2025FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Servingcited by 248
  8. 2020Ansor: Generating High-Performance Tensor Programs for Deep Learningcited by 199
  9. 2024NanoFlow: Towards Optimal Large Language Model Serving Throughputcited by 85
  10. 2024TorchTitan: One-stop PyTorch native solution for production ready LLM pre-trainingcited by 80
  11. 2021DNNFusion: Accelerating Deep Neural Networks Execution with Advanced Operator Fusioncited by 56
  12. 2022SparseTIR: Composable Abstractions for Sparse Compilation in Deep Learningcited by 47
9 more
  1. 2022TensorIR: An Abstraction for Automatic Tensorized Program Optimizationcited by 38
  2. 2025Triton-distributed: Programming Overlapping Kernels on Distributed AI Systems with the Triton Compilercited by 24
  3. 2020Automatic Horizontal Fusion for GPU Kernelscited by 17
  4. 2025MPK: A Compiler and Runtime for Mega-Kernelizing Tensor Programscited by 14
  5. 2025FlashMoE: Fast Distributed MoE in a Single Kernelcited by 13
  6. 2023Relax: Composable Abstractions for End-to-End Dynamic Machine Learningcited by 12
  7. 2023A Framework for Fine-Grained Synchronization of Dependent GPU Kernelscited by 5
  8. 2025Task-Based Tensor Computations on Modern GPUscited by 4
  9. 2026Axe: A Simple Unified Layout Abstraction for Machine Learning Compilerscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 4.

4 papers.

  1. 2026ExpertPlex: A High-Goodput Disaggregated Serving System for MoE LLMs with Adaptive Persistent Kernelscited by 1
  2. 2026PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloudcited by 1
  3. 2026Measuring and Reducing WebGPU Dispatch Overhead for LLM Inferencecited by 0
  4. 2026Ready Cohorts: Bounding GPU Opportunity and Avoiding Host Round Trips in LLM-Agent Controlcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026FastKernels: Benchmarking GPU Kernel Generation in Production
  2. 2026Parallel Track Transformers: Enabling Fast GPU Inference with Reduced Synchronization
  3. 2026TensorCast: The Missing Tensor Management Layer in Large Language Model Infrastructure
  4. 2026Automated Tensor Scheduling for Hybrid CPU-GPU LLM Inference on Consumer Devices
  5. 2026KernelSight-LM: A Kernel-Level LLM Inference Simulator
  6. 2026Scaling State-Space Models on Multiple GPUs with Tensor Parallelism
  7. 2025FailSafe: High-performance Resilient Serving
  8. 2025RAPID-LLM: Resilience-Aware Performance analysis of Infrastructure for Distributed LLM Training and Inference ↗