In modern healthcare, cloud-based e-health technology offers substantial benefits but faces significant security challenges. Sensitive patient data is vulnerable to cyber threats during transmission and storage, poten...
详细信息
The recent huge advance of Large Language Models (LLMs) is mainly driven by the increase in the number of parameters. This has led to substantial memory capacity requirements, necessitating the use of dozens of GPUs j...
The recent huge advance of Large Language Models (LLMs) is mainly driven by the increase in the number of parameters. This has led to substantial memory capacity requirements, necessitating the use of dozens of GPUs just to meet the capacity. One popular solution to this is storage-offloaded training, which uses host memory and storage as an extended memory hierarchy. However, this obviously comes at the cost of storage bandwidth bottleneck because storage devices have orders of magnitude lower bandwidth compared to that of GPU device memories. Our work, Smart-Infinity, addresses the storage bandwidth bottleneck of storage-offloaded LLM training using near-storage processing devices on a real system. The main component of Smart-Infinity is SmartUpdate, which performs parameter updates on custom near-storage accelerators. We identify that moving parameter updates to the storage side removes most of the storage traffic. In addition, we propose an efficient data transfer handler structure to address the system integration issues for Smart-Infinity. The handler allows overlapping data transfers with fixed memory consumption by reusing the device buffer. Lastly, we propose accelerator-assisted gradient compression/decompression to enhance the scalability of Smart-Infinity. When scaling to multiple near-storage processing devices, the write traffic on the shared channel becomes the bottleneck. To alleviate this, we compress the gradients on the GPU and decompress them on the accelerators. It provides further acceleration from reduced traffic. As a result, Smart-Infinity achieves a significant speedup compared to the baseline. Notably, SmartInfinity is a ready-to-use approach that is fully integrated into PyTorch on a real system. The implementation of Smart-Infinity is available at https://***/AIS-SNU/smart-infinity.
The study presented a novel chipless RFID sensor designed for the detection of cracks in non-metallic materials. The sensor utilized square coplanar split-ring resonators (SRR) and was mounted on the material under te...
详细信息
This article reports two important mixed-signal building blocks using unipolar oxide thin-film transistors (TFTs) on a 30- μ m -thick polymide substrate, which find potential application in communication systems. The...
详细信息
Far-feld chemical microscopy providing molecular electronic or vibrational fingerprint information opens a new window for the study of three-dimensional biological,material,and chemical *** microscopy provides a nonde...
详细信息
Far-feld chemical microscopy providing molecular electronic or vibrational fingerprint information opens a new window for the study of three-dimensional biological,material,and chemical *** microscopy provides a nondestructive way of chemical identification without exterior ***,the diffraction limit of optics hindered it from discovering more details under the resolution *** development of super-resolution techniques gives enlightenment to open this door behind far-field chemical ***,we review recent advances that have pushed the boundary of far-field chemical microscopy in terms of spatial *** further highlight applications in biomedical research,material characterization,environmental study,cultural heritage conservation,and integrated chip inspection.
Reversible data hiding is widely utilized for secure communication and copyright protection. Recently, to improve embedding capacity and visual quality of stego-images, some Partial Reversible Data Hiding (PRDH) schem...
详细信息
Non-volatile memory(NVM)provides a scalable and power-efficient solution to replace dynamic random access memory(DRAM)as main ***,because of the relatively high latency and low bandwidth of NVM,NVM is often paired wit...
详细信息
Non-volatile memory(NVM)provides a scalable and power-efficient solution to replace dynamic random access memory(DRAM)as main ***,because of the relatively high latency and low bandwidth of NVM,NVM is often paired with DRAM to build a heterogeneous memory system(HMS).As a result,data objects of the application must be carefully placed to NVM and DRAM for the best *** this paper,we introduce a lightweight runtime solution that automatically and transparently manages data placement on HMS without the requirement of hardware modifications and disruptive change to *** online profiling and performance models,the runtime solution characterizes memory access patterns associated with data objects,and minimizes unnecessary data *** runtime solution effectively bridges the performance gap between NVM and *** demonstrate that using NVM to replace the majority of DRAM can be a feasible solution for future HPC systems with the assistance of a software-based data management.
We present a bi-directional Fano resonator, which is capable of spectral shape control by adjusting the continuum state. By addressing the reflection phase, we demonstrate sensitive-chromatic response for color contro...
详细信息
The paper presents an innovative approach for predicting breast cancer by employing fine-needle aspiration (FNA), the synthetic minority oversampling method (SMOTE), and a Cubic Support Vector Machine (c-SVM) to gener...
详细信息
This study explores the relationship between quiz question performance and preceding quiz activities, aiming to predict the former through the lens of diverse machine learning models. Our research explores four distin...
详细信息
暂无评论