mlsanityData to 14 Oct 2026

Hardwired-Neurons Language Processing Units as General-Purpose Cognitive Substrates

arXiv 2508.16151 ↗First posted 22 Aug 2025cs.AR cs.CLcited by 1

Not on the map. Why?

Abstract (arXiv)

The rapid advancement of Large Language Models (LLMs) has established language as a core general-purpose cognitive substrate, driving the demand for specialized Language Processing Units (LPUs) tailored for LLM inference. To overcome the growing energy consumption of LLM inference systems, this paper proposes a Hardwired-Neurons Language Processing Unit (HNLPU), which physically hardwires LLM weight parameters into the computational fabric, achieving several orders of magnitude computational efficiency improvement by extreme specialization. However, a significant challenge still lies in the scale of modern LLMs. A straightforward hardwiring of gpt-oss 120 B would require fabricating photomask sets valued at over 6 billion dollars, rendering this straightforward solution economically impractical. Addressing this challenge, we propose the novel Metal-Embedding methodology. Instead of embedding weights in a 2D grid of silicon device cells, Metal-Embedding embeds weight parameters into the 3D topology of metal wires. This brings two benefits: (1) a 15x increase in density, and (2) 60 out of 70 photomask layers are homogeneous across chips, including all EUV photomasks. In total, Metal-Embedding reduced the photomask cost by 112x, bringing the Non-Recurring Engineering (NRE) cost of HNLPU into an economically viable range. Experimental results show that HNLPU achieved 249,960 tokens/s (5,555x/85x that of GPU/WSE), 36 tokens/J (1,047x/283x that of GPU/WSE), 13,232 mm2 total die area, 59.46 M-123.5 M estimated NRE at 5 nm technology. Analysis shows that HNLPU achieved 41.7-80.4x improvement in cost-effectiveness and 357x reduction in carbon footprint compared to OpenAI-scale H100 clusters, under an annual weight updating assumption.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 17 of its 98 references that are in the map. First posted on arXiv 22 Aug 2025.

Builds on

17 of 98 references.

References · 17

  1. 2017Attention Is All You Needcited by 60,361
  2. 2020Language Models are Few-Shot Learnerscited by 32,419
  3. 2024The Llama 3 Herd of Modelscited by 14,715
  4. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,653
  5. 2017Efficient Processing of Deep Neural Networks: A Tutorial and Surveycited by 434
  6. 2022DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scalecited by 407
  7. 2024Helix: Serving Large Language Models over Heterogeneous GPUs and Network via Max-Flowcited by 55
  8. 2025Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decodingcited by 55
  9. 2024ALISA: Accelerating Large Language Model Inference via Sparsity-Aware KV Cachingcited by 48
  10. 2025Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architecturescited by 39
  11. 2025TAPAS: Thermal- and Power-Aware Scheduling for LLM Inference in Cloud Platformscited by 29
  12. 2025PIM Is All You Need: A CXL-Enabled GPU-Free System for Large Language Model Inferencecited by 22
5 more
  1. 2025PAPI: Exploiting Dynamic Parallelism in Large Language Model Decoding with a Processing-In-Memory-Enabled Computing Systemcited by 20
  2. 2025EcoServe: Designing Carbon-Aware AI Inference Systemscited by 19
  3. 2025VQ-LLM: High-performance Code Generation for Vector Quantization Augmented LLM Inferencecited by 13
  4. 2024SOFA: A Compute-Memory Optimized Sparsity Accelerator via Cross-Stage Coordinated Tilingcited by 9
  5. 2025SoMa: Identifying, Exploring, and Understanding the DRAM Communication Scheduling Space for DNN Acceleratorscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Physical Foundation Models: Fixed hardware implementations of large-scale neural networkscited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2026BaseRT: Advancing Best-in-Class LLM Inference with Apple M5 Neural Accelerators
  2. 2025Make LLM Inference Affordable to Everyone: Augmenting GPU Memory with NDP-DIMM
  3. 2026BaseRT: Best-in-Class LLM Inference on Apple Silicon via Native Metal
  4. 2025SLIM: A Heterogeneous Accelerator for Edge Inference of Sparse Large Language Model via Adaptive Thresholding
  5. 2025Vec-LUT: Vector Table Lookup for Parallel Ultra-Low-Bit LLM Inference on Edge Devices
  6. 2026LLM Compression with Jointly Optimizing Architectural and Quantization choices
  7. 2025Pushing up to the Limit of Memory Bandwidth and Capacity Utilization for Efficient LLM Decoding on Embedded FPGA
  8. 2024LUT Tensor Core: A Software-Hardware Co-Design for LUT-Based Low-Bit LLM Inference