Session

· Contributors · Organizations · Search Program · My Agenda · Happening Now

Research Manuscript: Accelerating the Inference: Transformers, Graphs and Others

Session Chairs

Muhammad Shafique

New York University

Jeff (Jun) Zhang

Harvard University

Event TypeResearch Manuscript

Keywords

Topics

TimeWednesday, July 13th1:30pm - 3:00pm PDT

Location3002, Level 3

DescriptionThis session presents novel accelerators for neural inference. The papers presented target accelerators of graph neural networks and transformer networks among others. Further, one accelerator focuses on diagonal matrix multiplication, used in sparse self-attention heads. All papers present various trade-offs and demonstrate significant progress against the state of the art.

Presentations

1:30pm - 1:53pm PDT	GNNIE: GNN Inference Engine with Load-balancing and Graph-specific Caching Authors Sudipta Mondal Susmita Dey Manasi Kishor Kunal Ramprasath S Sachin S. Sapatnekar
1:53pm - 2:15pm PDT	SALO: an Efficient Spatial Accelerator Enabling Hybrid Sparse Attention Mechanisms for Long Sequences Authors Guan Shen Jieru Zhao Quan Chen Jingwen Leng Chao Li Minyi Guo
2:15pm - 2:37pm PDT	NN-LUT: Neural Approximation of Non-Linear Operations for Efficient Transformer Inference Authors Joonsang Yu Junki Park Seongmin Park Minsoo Kim Sihwa Lee Dong Hyun Lee Jungwook Choi
2:37pm - 3:00pm PDT	Self Adaptive Reconfigurable Arrays (SARA): Learning Flexible GEMM Accelerator Configuration and Mapping-space using ML Authors Ananda Samajdar Eric Qin Michael Pellauer Tushar Krishna