mlsanityData to 14 Oct 2026

LoopTune: Optimizing Tensor Computations with Reinforcement Learning

arXiv 2309.01825 ↗First posted 4 Sep 2023cs.LG cs.PLcited by 1

Not on the map. Why?

Abstract (arXiv)

Advanced compiler technology is crucial for enabling machine learning applications to run on novel hardware, but traditional compilers fail to deliver performance, popular auto-tuners have long search times and expert-optimized libraries introduce unsustainable costs. To address this, we developed LoopTune, a deep reinforcement learning compiler that optimizes tensor computations in deep learning models for the CPU. LoopTune optimizes tensor traversal order while using the ultra-fast lightweight code generator LoopNest to perform hardware-specific optimizations. With a novel graph-based representation and action space, LoopTune speeds up LoopNest by 3.2x, generating an order of magnitude faster code than TVM, 2.8x faster than MetaSchedule, and 1.08x faster than AutoTVM, consistently performing at the level of the hand-tuned library Numpy. Moreover, LoopTune tunes code in order of seconds.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 27 of its 71 references that are in the map. First posted on arXiv 4 Sep 2023.

Builds on

27 of 71 references.

References · 27

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  2. 2017Proximal Policy Optimization Algorithmscited by 15,215
  3. 2013Playing Atari with Deep Reinforcement Learningcited by 3,274
  4. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,258
  5. 2018IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architecturescited by 883
  6. 2017In-Datacenter Performance Analysis of a Tensor Processing Unitcited by 618
  7. 2018Distributed Prioritized Experience Replaycited by 269
  8. 2014cuDNN: Efficient Primitives for Deep Learningcited by 245
  9. 2020Ansor: Generating High-Performance Tensor Programs for Deep Learningcited by 201
  10. 2018Learning to Optimize Tensor Programscited by 140
  11. 2018Tensor Comprehensions: Framework-Agnostic High-Performance Machine Learning Abstractionscited by 85
  12. 2018NVIDIA Tensor Core Programmability, Performance & Precisioncited by 79
15 more
  1. 2018Glow: Graph Lowering Compiler Techniques for Neural Networkscited by 75
  2. 2019Dissecting the Graphcore IPU Architecture via Microbenchmarkingcited by 47
  3. 2020Action Space Shaping in Deep Reinforcement Learningcited by 46
  4. 2021CompilerGym: Robust, Performant Compiler Optimization Environments for AI Researchcited by 41
  5. 2018A Survey on Compiler Autotuning using Machine Learningcited by 40
  6. 2021MLGO: a Machine Learning Guided Compiler Optimizations Frameworkcited by 40
  7. 2019NeuroVectorizer: End-to-End Vectorization with Deep Reinforcement Learningcited by 37
  8. 2020Chameleon: Adaptive Code Optimization for Expedited Deep Neural Network Compilationcited by 29
  9. 2018Intel nGraph: An Intermediate Representation, Compiler, and Executor for Deep Learningcited by 23
  10. 2019Incidence Networks for Geometric Deep Learningcited by 17
  11. 2020Fast Stencil-Code Computation on a Wafer-Scale Processorcited by 10
  12. 2021A Reinforcement Learning Environment for Polyhedral Optimizationscited by 5
  13. 2016High-Performance Tensor Contraction without Transpositioncited by 3
  14. 2014A Tiling Perspective for Register Optimizationcited by 1
  15. 2022LoopStack: a Lightweight Tensor Algebra Compiler Stackcited by 1

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025A Deep Learning Model for Predicting Transformation Legalitycited by 0