General Reuse-Centric CNN Accelerator
نویسندگان
چکیده
This article introduces the first general reuse-centric accelerator for CNN inferences. Unlike prior work that exploits similarities only across consecutive video frames, is able to discover among arbitrary patches within an image or independent images, and translate them into computation time energy savings. Experiments show accelerator complements both software-based various hardware accelerators, producing up 14.96X speedups similarity discovery, 2.70X overall inference.
منابع مشابه
Design and Analysis of a Hardware CNN Accelerator
In recent years, Convolutional Neural Networks (CNNs) have revolutionized computer vision tasks. However, inference in current CNN designs is extremely computationally intensive. This has lead to an explosion of new accelerator architectures designed to reduce power consumption and latency [20]. In this paper, we design and implement a systolic array based architecture we call ConvAU to efficie...
متن کاملOn-Chip CNN Accelerator for Image Super-Resolution
To implement convolutional neural networks (CNN) in hardware, the state-of-the-art CNN accelerators pipeline computation and data transfer stages using an off-chip memory and simultaneously execute them on the same timeline. However, since a large amount of feature maps generated during the operation should be transmitted to the off-chip memory, the pipeline stage length is determined by the of...
متن کاملInventory Accelerator in General Equilibrium
We develop a general-equilibrium model of inventories with explicit micro-foundations by embedding the production-cost-smoothing motive (e.g., Eichenbaum, AER 1989) into an otherwise standard DSGE model. We show that rms facing idiosyncratic cost shocks have incentives to bunch production and smooth sales by carrying inventories. The optimal inventory target of a rm is derived explicitly. The...
متن کاملA Data-Reuse Aware Accelerator for Large-Scale Convolutional Networks
This paper presents a clustered SIMD accelerator template for Convolutional Networks. These networks significantly outperform other methods in detection and classification tasks in the vision domain. Due to the excessive compute and data transfer requirements these applications benefit a lot from a dedicated accelerator. The proposed accelerator reduces memory traffic by loop transformations su...
متن کاملIntegrating SAAM in Domain - centric and Reuse - based Development Processes *
Architecture analysis allows to assess an architecture with respect to various characteristics. For example, it is possible to predict non-functional properties of applications based on the architecture , and to estimate the effort for specific maintenance situations. After a short introduction to the SAAM method for scenario-based architecture analysis, we point out how this technique can be d...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE Transactions on Computers
سال: 2022
ISSN: ['1557-9956', '2326-3814', '0018-9340']
DOI: https://doi.org/10.1109/tc.2021.3064608