Guardado en:
Detalles Bibliográficos
Autores principales: Moon, Seunghyun, Li, Mao, Chen, Gregory, Knag, Phil, Krishnamurthy, Ram, Seok, Mingoo
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2503.00322
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917941196357632
author Moon, Seunghyun
Li, Mao
Chen, Gregory
Knag, Phil
Krishnamurthy, Ram
Seok, Mingoo
author_facet Moon, Seunghyun
Li, Mao
Chen, Gregory
Knag, Phil
Krishnamurthy, Ram
Seok, Mingoo
contents This work introduces novel training and post-training compression schemes to reduce external memory access during transformer model inference. Additionally, a new control flow mechanism, called dynamic batching, and a novel buffer architecture, termed a two-direction accessible register file, further reduce external memory access while improving hardware utilization.
format Preprint
id arxiv_https___arxiv_org_abs_2503_00322
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle T-REX: A 68-567 μs/token, 0.41-3.95 μJ/token Transformer Accelerator with Reduced External Memory Access and Enhanced Hardware Utilization in 16nm FinFET
Moon, Seunghyun
Li, Mao
Chen, Gregory
Knag, Phil
Krishnamurthy, Ram
Seok, Mingoo
Hardware Architecture
Artificial Intelligence
This work introduces novel training and post-training compression schemes to reduce external memory access during transformer model inference. Additionally, a new control flow mechanism, called dynamic batching, and a novel buffer architecture, termed a two-direction accessible register file, further reduce external memory access while improving hardware utilization.
title T-REX: A 68-567 μs/token, 0.41-3.95 μJ/token Transformer Accelerator with Reduced External Memory Access and Enhanced Hardware Utilization in 16nm FinFET
topic Hardware Architecture
Artificial Intelligence
url https://arxiv.org/abs/2503.00322