mlsanityData to 14 Oct 2026

Ember: A Compiler for Efficient Embedding Operations on Decoupled Access-Execute Architectures

arXiv 2504.09870 ↗First posted 14 Apr 2025cs.AR cs.LG cs.PLcited by 0

Not on the map. Why?

Abstract (arXiv)

Irregular embedding lookups are a critical bottleneck in recommender models, sparse large language models, and graph learning models. In this paper, we first demonstrate that, by offloading these lookups to specialized access units, Decoupled Access-Execute (DAE) processors achieve 2.6× higher performance and 6.4× higher performance/watt than GPUs on end-to-end models. Then, we propose the Ember compiler for automatically generating optimized DAE code from PyTorch and TensorFlow. Conversely from other DAE compilers, Ember features multiple intermediate representations specifically designed for different optimization levels. In this way, Ember can implement all optimizations to match the performance of hand-written code, unlocking the full potential of DAE architectures at scale.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 22 of its 65 references that are in the map. First posted on arXiv 14 Apr 2025.

Builds on

22 of 65 references.

References · 22

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  3. 2017Inductive Representation Learning on Large Graphscited by 6,320
  4. 2019Fast Graph Representation Learning with PyTorch Geometriccited by 2,434
  5. 2020Big Bird: Transformers for Longer Sequencescited by 1,684
  6. 2014Caffe: Convolutional Architecture for Fast Feature Embeddingcited by 288
  7. 2019The Architectural Implications of Facebook's DNN-based Personalized Recommendationcited by 83
  8. 2019RecNMP: Accelerating Personalized Recommendation with Near-Memory Processingcited by 49
  9. 2022SparseTIR: Composable Abstractions for Sparse Compilation in Deep Learningcited by 47
  10. 2020Understanding Training Efficiency of Deep Learning Recommendation Models at Scalecited by 43
  11. 2019Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systemscited by 41
  12. 2019Neural Input Search for Large Scale Recommendation Modelscited by 39
10 more
  1. 2022RecShard: Statistical Feature-Based Memory Optimization for Industry-Scale Neural Recommendationcited by 28
  2. 2020FusedMM: A Unified SDDMM-SpMM Kernel for Graph Embedding and Graph Neural Networkscited by 26
  3. 2020The gem5 Simulator: Version 20.0+cited by 14
  4. 2018The ARM Scalable Vector Extensioncited by 9
  5. 2022Looplets: A Language For Structured Coiterationcited by 4
  6. 2024UniSparse: An Intermediate Language for General Sparse Format Customizationcited by 4
  7. 2024Compilation of Modular and General Sparse Workspacescited by 4
  8. 2021Scaling up HBM Efficiency of Top-K SpMV for Approximate Embedding Similarity on FPGAscited by 1
  9. 2021Solving Large Top-K Graph Eigenproblems with a Memory and Compute-optimized FPGA Designcited by 1
  10. 2023SpChar: Characterizing the Sparse Puzzle via Decision Treescited by 1

Built on by

Not cited yet.