A 28nm 27.5TOPS/W Approximate-Computing-Based Transformer Processor with Asymptotic Sparsity Speculating and Out-of-Order Computing-Reference-Cited by-同舟云学术

A 28nm 27.5TOPS/W Approximate-Computing-Based Transformer Processor with Asymptotic Sparsity Speculating and Out-of-Order Computing

Published:2022-02-20 Issue: Volume: Page:
ISSN:
Container-title:2022 IEEE International Solid- State Circuits Conference (ISSCC)
language:
Short-container-title:

Author:

Wang Yang¹,Qin Yubin¹,Deng Dazheng¹,Wei Jingchuan¹,Zhou Yang¹,Fan Yuanqi¹,Chen Tianbao²,Sun Hao¹,Liu Leibo¹,Wei Shaojun¹,Yin Shouyi¹

Affiliation:

1. Tsinghua University,Beijing,China

2. Tsing Micro,Beijing,China

Funder

NSFC

National Key Research and Development Program

Beijing Innovation Center for Future Chip

BAAl

Publisher

IEEE

Link

http://xplorestaging.ieee.org/ielx7/9731529/9731102/09731686.pdf?arnumber=9731686

Reference6 articles.

1. An 8.93 TOPS/W LSTM Recurrent Neural Network Accelerator Featuring Hierarchical Coarse-Grain Sparsity for On-Device Speech Recognition

2. ELSA: Hardware-Software Co-design for Efficient, Lightweight Self-Attention Mechanism in Neural Networks

3. NVIDIA A100 Tensor Core GPU Architecture;nVidia Whitepaper,0

4. A 25mm2 SoC for loT Devices with 18ms Noise-Robust Speech-to-Text Latency via Bayesian Speech Denoising and Attention-Based Sequence-to-Sequence DNN Speech Recognition in 16nm FinFET;tambe;ISSCC,2021

5. A3: Accelerating Attention Mechanisms in Neural Networks with Approximation;ham;IEEE HPCA,2020

Cited by 17 articles. 订阅此论文施引文献订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献

1. HARDSEA: Hybrid Analog-ReRAM Clustering and Digital-SRAM In-Memory Computing Accelerator for Dynamic Sparse Self-Attention in Transformer;IEEE Transactions on Very Large Scale Integration (VLSI) Systems;2024-02

2. MulTCIM: Digital Computing-in-Memory-Based Multimodal Transformer Accelerator With Attention-Token-Bit Hybrid Sparsity;IEEE Journal of Solid-State Circuits;2024-01

3. P³ ViT: A CIM-Based High-Utilization Architecture With Dynamic Pruning and Two-Way Ping-Pong Macro for Vision Transformer;IEEE Transactions on Circuits and Systems I: Regular Papers;2023-12

4. To Spike or Not to Spike: A Digital Hardware Perspective on Deep Learning Acceleration;IEEE Journal on Emerging and Selected Topics in Circuits and Systems;2023-12

5. A 28nm 49.7TOPS/W Sparse Transformer Processor with Random-Projection-Based Speculation, Multi-Stationary Dataflow, and Redundant Partial Product Elimination;2023 IEEE Asian Solid-State Circuits Conference (A-SSCC);2023-11-05