Yisong Lin's research while affiliated with The Logistics Institute, Northeastern University and other places

Publications (17)

Article
Full-text available
Cloud storage system provides reliable service to users by widely deploying redundancy schemes in its system – which brings high reliability to the data storage, but inversely introduces significant overhead to the system, consisting of storage cost and energy consumption. The core behind this issue is how to leverage the relationship between data...
Article
Full-text available
In modern supply chain management systems, Radio Frequency IDentification (RFID) technology has become an indispensable sensor technology and massive RFID data sets are expected to become commonplace. More and more space and time are needed to store and process such huge amounts of RFID data, and there is an increasing realization that the existing...
Conference Paper
Reducing service cost has been a popular topic in recent studies on cloud storage systems. One of the basic techniques is to power down parts of service nodes. However, it will reduce the availability of data objects, which is the overriding concern of users. So we mathematically formulate the problem to close maximum service nodes under the constr...
Article
Full-text available
Radio Frequency IDentification (RFID) technology promises to revolutionize the way we track items and assets, but in RFID systems, missreading is a common phenomenon and it poses an enormous challenge to RFID data management, so accurate data cleaning becomes an essential task for the successful deployment of systems. In this paper, we present the...
Article
High performance computing can be well supported by the Grid or cloud computing systems. However, these systems have to overcome the failure risks, where data is stored in the “unreliable” storage nodes that can leave the system at any moment and the nodes’ network bandwidth is limited. In this case, the basic way to assure data reliability is to a...
Article
Recently, GPU has been widely used in High Performance Computing (HPC). In order to improve computational performance, several GPUs are integrated into one computer node in practical system. However, power consumption of GPUs is very high and becomes as bottleneck to its further development. In doing so, optimizing power consumption have been draw...
Article
Power management has become one of the first-order considerations in high performance computing field. Many recent studies focus on optimizing the performance of a computer system within a given power budget. However, most existing solutions adopt fixed period control mechanism and are transparent to the running applications. Although the applicati...
Article
GPUs render higher computing unit density than contemporary CPUs and thus exhibit much higher power consumption despite its higher power efficiency. The power consumption has become an important issue that impacts GPU's applications, thereby necessitating the low power optimization technology for GPUs. Software prefetching is an efficient way to al...
Conference Paper
Using the graphics processing unit (GPU) to accelerate the general purpose computation has attracted much attention from both the academia and industry due to GPU's powerful computing capacity. Thus optimization of GPU programs has become a popular research direction. In order to support the general purpose computing more efficiently, GPU has integ...
Article
As one of the most popular accelerators, Graphics Processing Unit (GPU) has demonstrated high computing power in several application fields. On the other hand, GPU also produces high power consumption and has been one of the most largest power consumers in desktop and supercomputer systems. However, software power optimization method targeted for G...
Article
Network calculus is a promising theory for analyzing and modeling networks based on min-plus algebra. Using network calculus theory, we propose formulas of arrival curve and service curve for end-to-end communication, build the corresponding time model, and derive the communication delay formulas for two scenarios of the model respectively. Then we...
Conference Paper
In light of its powerful computing capacity and high energy efficiency, GPU (graphics processing unit) has become a focus in the research field of HPC (High Performance Computing). CPU-GPU heterogeneous parallel systems have become a new development trend of super-computer. However, the inherent unreliability of the GPU hardware deteriorates the re...
Article
With the growth of supercomputer's scale, the communication time during executing is increasing. This phenomenon arouses the architecture researchers' interests. In this paper, based on the fat-tree topology, which is widely used in Infiniband, we present an one-to-all broadcast communication time model. After classifying applications into two kind...
Article
OpenMP is a widely used parallel programming model on traditional multi-core processors. Generally, OpenMP is used to develop fine-grained parallelism through a multi-thread model. Stream programming model is a new kind of parallel programming model for stream architectures. OpenMP bears a resemblance to the stream programming model at some level....
Conference Paper
In recent years, heterogeneous parallel system have become a focus research area in high performance computing field. Generally, in a heterogeneous parallel system, CPU provides the basic computing environment and special purpose accelerator (GPU in this paper) provides high computing performance. However, the overall performance of the system is p...
Conference Paper
In a chip-multiprocessor with a shared cache structure, the last level cache is shared by multiple applications executing simultaneously. The competing accesses from different applications degrade the system performance, resulting in non-predicting executing time. Cache partitioning techniques partition the shared cache for multiple applications. T...

Citations

... Indeed, the Grid application to the municipality scale can help understand power balances among municipalities, thus providing an in-depth knowledge of its dynamics. On the other hand, it is oriented to reduce the Grid's redundancy, intended as the presence of more than one indicator providing the same piece of information (Huang et al. 2015). More in detail, this task mainly addresses temporal redundancy and, thus, rejects indicators occurring twice with different time horizons when their simultaneous presence doesn't pitch in understanding the ongoing territorial dynamics (Table 2). ...
... Currently, the most widely used distributed cloud storage systems are Google File System (GFS) [13,25] and Hadoop Distributed File System (HDSF) [26,27]. In GFS, there exists a master server which stores three types of metadata: the file and chunk namespaces, the mapping from files to chunks and the locations of each chunk's replicas [13]. ...
... By contrast, erasure code (e.g., Reed-Solomon [13]) encodes k data blocks into ðn À kÞ coded blocks, resulting in n blocks in total. These n blocks are then distributed into different nodes [14,15]. ...
... Greengpu [13] involves low level programming and memory management with custom pthreadbased kernel launches for the GPU to divide workload between CPU and GPU for synthetic benchmarks. The authors in [14] utilize software prefetching and DVFS to reduce GPU energy consumption. ...
... Kernel fusion, which combines two kernels into a single thread, is proposed in [12] to improve GPU utilization and reduce energy consumption. Greengpu [13] involves low level programming and memory management with custom pthreadbased kernel launches for the GPU to divide workload between CPU and GPU for synthetic benchmarks. ...
... There is a drive for increased speed in processes using RFID portable devices in construction supply chains [78]. The supply chain can generate a vast amount of data that need to be processed and managed [79] using an efficient storage scheme and query process on relational databases [80,81], no-relational repository designs in MongoDB [82], or a method for processing massive amounts of data using path encoding [83]. All these data can be either important or not, based on their attributes [84]. ...
... Existing data cleaning studies on RFID and sensor mainly focus on proposing noise and invalid data processing methods. By summarizing the existing methods [Chen, Yu, Gu et al. (2011);Jiang, Xiao, Wang et al. (2011);Ziekow, Ivantysynova and Günter (2011);Ali, Pissinou and Makki (2012);Fan, Wu and Lin (2012); Zhang, Kalasapudi and Tang (2016)], we find that the research in this area can still be improved in terms of cleaning accuracy. This research will achieve excellent results for RFID and sensors to adjust data business logic according to the data cleaning result. ...
... Stack distance analysis in GPU kernels Wang and Xiao [42] showed the usefulness of SDA in GPU cache performance analysis. The first SDA method for analyzing GPU applications is proposed by Tand et al. [39]. The authors analyze CUDA blocks through stack distance and only modeled the conflicting effects of running CUDA blocks on SMs, ignoring the details of the GPU execution model. ...
... HPC community has developed various solutions to generally tolerate faults, and more specifically to mitigate faults caused by hardware defects [10] and to detect and recover from errors [5,11]. We will elaborate more on some of the relevant approaches in Section 2. Some of the used approaches depend on using checkpoints/reset [12], redundancy and Algorithm-Based Fault Tolerance ABFT [13,14]. In our research, we have applied redundancy-based fault tolerance, as checkpointing has high communication overhead and ABFT is customized to fit the algorithm under analysis, thus, it is very difficult to generalize the solution to other applications without addressing the specifics of the new algorithm. ...
... IPC metric for throughput: Though a significant amount of work is done using miss rate, above studies reveal that alone miss rate based cache partitioning proves to be sub-optimal when system performance is estimated using IPC metric. Literature reveals (Suo et al., 2008;Subramanian et al., 2015) that reducing miss rate does not necessarily mean increase in throughput, while increase in IPC could be a direct metric to measure throughput. ...