paper-with-me

홈 › Papers

HH-PIM: Dynamic Optimization of Power and Performance with Heterogeneous-Hybrid PIM for Edge AI Devices

2025-04-02 · Sangmin Jeon, Kangju Lee, Kyeongwon Lee, Woojoo Lee

Processing-in-Memory (PIM) architectures offer promising solutions for efficiently handling AI applications in energy-constrained edge environments. While traditional PIM designs enhance performance and energy efficiency by reducing data movement between memory and processing units, they are limited in edge devices due to continuous power demands and the storage requirements of large neural network weights in SRAM and DRAM. Hybrid PIM architectures, incorporating non-volatile memories like MRAM and ReRAM, mitigate these limitations but struggle with a mismatch between fixed computing resources and dynamically changing inference workloads. To address these challenges, this study introduces a Heterogeneous-Hybrid PIM (HH-PIM) architecture, comprising high-performance MRAM-SRAM PIM modules and low-power MRAM-SRAM PIM modules. We further propose a data placement optimization algorithm that dynamically allocates data based on computational demand, maximizing energy efficiency. FPGA prototyping and power simulations with processors featuring HH-PIM and other PIM types demonstrate that the proposed HH-PIM achieves up to $60.43$ percent average energy savings over conventional PIMs while meeting application latency requirements. These results confirm the suitability of HH-PIM for adaptive, energy-efficient AI processing in edge devices.

📄 PDF Abstract BibTeX arXiv:2504.01468

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Hybrid NOMA Assisted Heterogeneous Semantic and Bit Users Communication

2025-05-06 · Ishtiaque Ahmed, Leila Musavian

In this paper, we utilize a downlink hybrid Non-Orthogonal Multiple Access (NOMA) framework to support multiple semantic and bit users within the communication network. The hybrid NOMA setup exploits both NOMA and Orthog…

Neural Architecture Search of Hybrid Models for NPU-CIM Heterogeneous AR/VR Devices

2024-10-10 · Yiwei Zhao, Ziyun Li, Win-San Khwa, Xiaoyu Sun 외

Low-Latency and Low-Power Edge AI is essential for Virtual Reality and Augmented Reality applications. Recent advances show that hybrid models, combining convolution layers (CNN) and transformers (ViT), often achieve sup…

Neural Architecture Search

Hydra: Hybrid Server Power Model

2022-07-20 · Nigel Bernard, Hoa Nguyen, Aman Chandan, Savyasachi Jagdeeshan 외

With the growing complexity of big data workloads that require abundant data and computation, data centers consume a tremendous amount of power daily. In an effort to minimize data center power consumption, several studi…

modelScheduling

Hybrid Parallel Collaborative Simulation Framework Integrating Device Physics with Circuit Dynamics for PDAE-Modeled Power Electronic Equipment

2025-01-17 · Qingyuan Shi, Chijie Zhuang, Jiapeng Liu, Bo Lin 외

Optimizing high-performance power electronic equipment, such as power converters, requires multiscale simulations that incorporate the physics of power semiconductor devices and the dynamics of other circuit components, …

Blocking

Grid-Forming Control of Modular Dynamic Virtual Power Plants

2024-10-18 · Xiuqiang He, Josué Duarte, Verena Häberle, Florian Dörfler

This article explores a flexible and coordinated control design for an aggregation of heterogeneous distributed energy resources (DERs) in a dynamic virtual power plant (DVPP). The control design aims to provide a desire…