0% Complete
English
صفحه اصلی
/
پانزدهمین کنفرانس بین المللی فناوری اطلاعات و دانش
Architectural Insights: Comparing Weight Stationary and Output Stationary Systolic Arrays for Efficient Computation
نویسندگان :
Mahdi Kalbasi
1
1- University of Isfahan
کلمات کلیدی :
Systolic arrays،Convolutional Neural Networks،Output stationary،Weight stationary
چکیده :
This paper compares two prevalent architectures in systolic arrays: weight stationary and output stationary methods. Systolic arrays utilize interconnected processing elements (PEs) to perform parallel processing, making them suitable for applications in digital signal processing, image processing, and machine learning. We focus on their implementation of 2D matrix multiplication, a fundamental operation in neural networks. Simulations were conducted using Verilog HDL within the Xilinx Vivado Design Suite 2019, employing a 3x1 input matrix and a 3x3 weight matrix. Results confirmed the functionality of both architectures, with output matrices matching expected results. Weight stationary designs minimized data movement, while output stationary designs enhanced throughput through effective input data reuse. With a critical path delay of approximately 8.8 ns, corresponding to a maximum frequency of about 113 MHz, the study highlights that the critical path remains stable when scaling the number of PEs. Overall, this research validates the effectiveness of both architectures in high-performance matrix operations, offering valuable insights for future systolic array designs.
لیست مقالات
لیست مقالات بایگانی شده
ISAAF: بهبود چارچوب مجوز خودتطبیق SAAF با استفاده از پیادهسازی مبتنی بر عامل و مفهوم I-Shairing
الهام معین الدینی - دکتر منیره عبدوس - دکتر اسلام ناظمی
A Survey on Utilizing Reinforcement Learning in Wireless Sensor Networks Routing Protocols
Ali Forghani Elah Abadi - Seyedeh Elham Asghari - Sepideh Sharifani - Seyyed Amir Asghari - Mohammadreza Binesh Marvasti
Optimal selection of seed nodes by reducing the influence of common nodes in the influence maximization problem
Farzaneh Kazemzadeh - Ali Asghar Safaei - Mitra Mirzarezaee
Knowledge Graph Based Retrieval-Augmented Generation for Multi-Hop Question Answering Enhancement
Mahdi Amiri Shavaki - Pouria Omrani - Ramin Toosi - Mohammad Ali Akhaee
پیاده سازی موازی یک طرح (t,n)-تسهیم چند تصویر با استفاده از GPU
سعیده کبیری راد
Fast Online Character Recognition Using a Novel Local-Global Feature Extraction Method
Ayoub Parvizi - Dr Mohammad Kazemifard - Ziba Imani
A Topic Based Method to Classify the Question Clarity in CQA Networks
Alireza Khabbazan - Dr Ahmad Ali Abin
ارائه یک رویکرد معنایی مبتنی بر آنتولوژی به منظور شناسایی تاکتیکهای معماری
احسان شریفی - دکتر احمد عبدالله زاده بارفروش
GanjNet: Leveraging Network Modeling with Large Language Models for Persian Word Sense Induction
Amir Mohammad Kouyeshpour - Hadi Veisi - Saman Haratizadeh
A Neural-based Approach to Aid Early Parkinson's Disease Diagnosis
Dr Armin Salimi-badr - Mohammad Hashemi
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 41.3.1