mlsanityData to 14 Oct 2026

Stella Nera: A Differentiable Maddness-Based Hardware Accelerator for Efficient Approximate Matrix Multiplication

arXiv 2311.10207 ↗First posted 16 Nov 2023cs.AR cs.CV cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Artificial intelligence has surged in recent years, with advancements in machine learning rapidly impacting nearly every area of life. However, the growing complexity of these models has far outpaced advancements in available hardware accelerators, leading to significant computational and energy demands, primarily due to matrix multiplications, which dominate the compute workload. Maddness (i.e., Multiply-ADDitioN-lESS) presents a hash-based version of product quantization, which renders matrix multiplications into lookups and additions, eliminating the need for multipliers entirely. We present Stella Nera, the first Maddness-based accelerator achieving an energy efficiency of 161 TOp/s/[email protected], 25x better than conventional MatMul accelerators due to its small components and reduced computational complexity. We further enhance Maddness with a differentiable approximation, allowing for gradient-based fine-tuning and achieving an end-to-end performance of 92.5% Top-1 accuracy on CIFAR-10.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 20 of its 35 references that are in the map. First posted on arXiv 16 Nov 2023.

Builds on

20 of 35 references.

References · 20

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,079
  3. 2015Compressing Neural Networks with the Hashing Trickcited by 150
  4. 2019SLIDE : In Defense of Smart Algorithms over Hardware Acceleration for Large-Scale Deep Learning Systemscited by 46
  5. 2019Differentiable Product Quantization for End-to-End Embedding Compressioncited by 36
  6. 2021DeepScaleTool : A Tool for the Accurate Estimation of Technology Scaling in the Deep-Submicron Eracited by 30
  7. 2021Multiplying Matrices Without Multiplyingcited by 30
  8. 2022AI and ML Accelerator Survey and Trendscited by 23
  9. 2023LUT-NN: Empower Efficient Neural Network Inference with Centroid Learning and Table Lookupcited by 20
  10. 2016Scalable and Sustainable Deep Learning via Randomized Hashingcited by 17
  11. 2020CUTIE: Beyond PetaOp/s/W Ternary DNN Inference Acceleration with Better-than-Binary Energy Efficiencycited by 10
  12. 2020ChewBaccaNN: A Flexible 223 TOPS/W BNN Acceleratorcited by 9
8 more
  1. 2023Flex-SFU: Accelerating DNN Activation Functions by Non-Uniform Piecewise Approximationcited by 8
  2. 2017Bolt: Accelerated Data Mining with Fast Vector Compressioncited by 7
  3. 2020FPnew: An Open-Source Multi-Format Floating-Point Unit Architecture for Energy-Proportional Transprecision Computingcited by 7
  4. 2022RedMulE: A Compact FP16 Matrix-Multiplication Accelerator for Adaptive Deep Learning on RISC-V-Based Ultra-Low-Power SoCscited by 7
  5. 2022PECAN: A Product-Quantized Content Addressable Memory Networkcited by 6
  6. 2022Look-ups are not (yet) all you need for deep learning inferencecited by 4
  7. 2021Yet Another Representation of Binary Decision Trees: A Mathematical Demonstrationcited by 2
  8. 2017Design Automation for Binarized Neural Networks: A Quantum Leap Opportunity?cited by 1

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025SAIL: SRAM-Accelerated LLM Inference System with Lookup-Table-based GEMVcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2022A Soft SIMD Based Energy Efficient Computing Microarchitecture
  2. 2026W4A16 Mixed-Precision Matrix Multiplication on Decoupled Architecture: Kernel Design and Memory Bottleneck Analysis for Ascend NPUs
  3. 2025The Cambrian Explosion of Mixed-Precision Matrix Multiplication for Quantized Deep Learning Inference
  4. 2025Accelerating Sparse Ternary GEMM for Quantized ML on Apple Silicon
  5. 2025Platinum: Path-Adaptable LUT-Based Accelerator Tailored for Low-Bit Weight Matrix Multiplication
  6. 2021A Deep Learning Inference Scheme Based on Pipelined Matrix Multiplication Acceleration Design and Non-uniform Quantization
  7. 2024Accelerating a Triton Fused Kernel for W4A16 Quantized Inference with SplitK work decomposition
  8. 2025Transitive Array: An Efficient GEMM Accelerator with Result Reuse