mlsanityData to 14 Oct 2026

SEADA: An efficient methodology for optimizing mixed-precision DNNs on multi-precision spatial architectures

arXiv 2606.27884 ↗First posted 26 Jun 2026cs.AR cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Mixed-precision computation has been introduced in deep neural networks (DNNs) as an effective approach to reduce latency, energy consumption, and memory footprint. However, efficiently mapping mixed-precision networks onto multi-precision spatial architectures poses several challenges. These include determining the appropriate precision for each layer, balancing layer-wise accuracy sensitivity to quantization against architectural heterogeneity and system-level constraints, and accurately estimating the system-level cost of heterogeneous precision assignments. This work presents SEADA, an efficient methodology designed to address these challenges. SEADA comprises: (i) a configurable system-level analytical cost model of a multi-precision spatial accelerator architecture; (ii) a fast mapping tool that identifies near-optimal mappings of DNN workloads onto the target integer accelerator; (iii) analytical models for floating-point layers to estimate the overall benefits of mixed-precision execution; and (iv) a per-layer precision selection methodology based on bit-level entropy, enabling efficient assignment across multiple numerical precisions. SEADA's efficiency provides designers with a robust framework for the design-space exploration of multi-precision architectures.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 13 of its 36 references that are in the map. First posted on arXiv 26 Jun 2026.

Builds on

13 of 36 references.

References · 13

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  4. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,148
  5. 2021A Survey of Quantization Methods for Efficient Neural Network Inferencecited by 815
  6. 2016DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradientscited by 569
  7. 2021I-BERT: Integer-only BERT Quantizationcited by 255
  8. 2024QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Servingcited by 208
  9. 2022FP8 Quantization: The Power of the Exponentcited by 88
  10. 2021Towards Mixed-Precision Quantization of Neural Networks via Constrained Optimizationcited by 61
  11. 2024A Stochastic Rounding-Enabled Low-Precision Floating-Point MAC for DNN Trainingcited by 6
  12. 2023Efficient and Effective Methods for Mixed Precision Neural Network Quantization for Faster, Energy-efficient Inferencecited by 3
1 more
  1. 2023SALSA: Simulated Annealing based Loop-Ordering Scheduler for DNN Acceleratorscited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2021Rethinking Floating Point Overheads for Mixed Precision DNN Accelerators
  2. 2024Exploring Quantization and Mapping Synergy in Hardware-Aware Deep Neural Network Accelerators
  3. 2022Mixed-Precision Neural Networks: A Survey
  4. 2022Channel-wise Mixed-precision Assignment for DNN Inference on Constrained Edge Nodes
  5. 2022Design of High-Throughput Mixed-Precision CNN Accelerators on FPGA
  6. 2023Free Bits: Latency Optimization of Mixed-Precision Quantized Neural Networks on the Edge
  7. 2024Mixed-precision Neural Networks on RISC-V Cores: ISA extensions for Multi-Pumped Soft SIMD Operations
  8. 2023LRMP: Layer Replication with Mixed Precision for Spatial In-memory DNN Accelerators