StreamDCIM: A Tile-based Streaming Digital CIM Accelerator with Mixed-stationary Cross-forwarding Dataflow for Multimodal Transformer

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Qin, Shantian, Qiang, Ziqing, Fan, Zhihua, Li, Wenming, An, Xuejun, Ye, Xiaochun, Fan, Dongrui
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916605743595520
author Qin, Shantian
Qiang, Ziqing
Fan, Zhihua
Li, Wenming
An, Xuejun
Ye, Xiaochun
Fan, Dongrui
author_facet Qin, Shantian
Qiang, Ziqing
Fan, Zhihua
Li, Wenming
An, Xuejun
Ye, Xiaochun
Fan, Dongrui
contents Multimodal Transformers are emerging artificial intelligence (AI) models designed to process a mixture of signals from diverse modalities. Digital computing-in-memory (CIM) architectures are considered promising for achieving high efficiency while maintaining high accuracy. However, current digital CIM-based accelerators exhibit inflexibility in microarchitecture, dataflow, and pipeline to effectively accelerate multimodal Transformer. In this paper, we propose StreamDCIM, a tile-based streaming digital CIM accelerator for multimodal Transformers. It overcomes the above challenges with three features: First, we present a tile-based reconfigurable CIM macro microarchitecture with normal and hybrid reconfigurable modes to improve intra-macro CIM utilization. Second, we implement a mixed-stationary cross-forwarding dataflow with tile-based execution decoupling to exploit tile-level computation parallelism. Third, we introduce a ping-pong-like fine-grained compute-rewriting pipeline to overlap high-latency on-chip CIM rewriting. Experimental results show that StreamDCIM outperforms non-streaming and layer-based streaming CIM-based solutions by geomean 2.63$\times$ and 1.28$\times$ on typical multimodal Transformer models.
format Preprint
id arxiv_https___arxiv_org_abs_2502_05798
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle StreamDCIM: A Tile-based Streaming Digital CIM Accelerator with Mixed-stationary Cross-forwarding Dataflow for Multimodal Transformer
Qin, Shantian
Qiang, Ziqing
Fan, Zhihua
Li, Wenming
An, Xuejun
Ye, Xiaochun
Fan, Dongrui
Hardware Architecture
Multimodal Transformers are emerging artificial intelligence (AI) models designed to process a mixture of signals from diverse modalities. Digital computing-in-memory (CIM) architectures are considered promising for achieving high efficiency while maintaining high accuracy. However, current digital CIM-based accelerators exhibit inflexibility in microarchitecture, dataflow, and pipeline to effectively accelerate multimodal Transformer. In this paper, we propose StreamDCIM, a tile-based streaming digital CIM accelerator for multimodal Transformers. It overcomes the above challenges with three features: First, we present a tile-based reconfigurable CIM macro microarchitecture with normal and hybrid reconfigurable modes to improve intra-macro CIM utilization. Second, we implement a mixed-stationary cross-forwarding dataflow with tile-based execution decoupling to exploit tile-level computation parallelism. Third, we introduce a ping-pong-like fine-grained compute-rewriting pipeline to overlap high-latency on-chip CIM rewriting. Experimental results show that StreamDCIM outperforms non-streaming and layer-based streaming CIM-based solutions by geomean 2.63$\times$ and 1.28$\times$ on typical multimodal Transformer models.
title StreamDCIM: A Tile-based Streaming Digital CIM Accelerator with Mixed-stationary Cross-forwarding Dataflow for Multimodal Transformer
topic Hardware Architecture
url https://arxiv.org/abs/2502.05798