Filters








29,468 Hits in 8.8 sec

Simulation studies of a virtual memory, time-shared, demand paging operating system

J. Winograd, S. J. Morganstein, R. Herman
1971 Proceedings of the third symposium on Operating systems principles - SOSP '71  
SIM/61 is a large (3500 lines of Simscript code), highly detailed simulation model of a virtual memory, time-shared, demand paging operating system.  ...  The approach taken for the former was to establish a fixed load, and measure its performance on various hardware configurations.  ...  Oppenheimer of RCA, who supplied the original impetus and direction for the backing store study. Deep appreciation goes to M. Fogel who has spent many long hours with us analyzing simulation output.  ... 
doi:10.1145/800212.806513 dblp:conf/sosp/WinogradMH71 fatcat:pm5354p7h5aypf6gako6rrnxoq

Design and Analysis of Benchmarking Experiments for Distributed Internet Services

Eytan Bakshy, Eitan Frachtenberg
2015 Proceedings of the 24th International Conference on World Wide Web - WWW '15  
Even small regressions in processing time can translate directly into significant energy and user experience costs.  ...  Using simulations and empirical data from Perflab, we validate our theoretical results, and provide easy-to-implement guidelines for designing and analyzing such benchmarks.  ...  users) to a service, and measuring the sample averageȳ, for some outcome yi for each observation.  ... 
doi:10.1145/2736277.2741082 dblp:conf/www/BakshyF15 fatcat:awhify27g5ax7oyrisywy74gwu

Algorithms and Architecture for Real-Time Recommendations at News UK [chapter]

Dion Bailey, Tom Pajak, Daoud Clarke, Carlos Rodriguez
2017 Lecture Notes in Computer Science  
In this paper we describe a new algorithm for updating collaborative filtering models incrementally, and demonstrate its effectiveness on clickstream data from The Times.  ...  However, little has been published about systems that can generate recommendations in response to changes in recommendable items and user behaviour in a very short space of time.  ...  Acknowledgments Many thanks to Dan Gilbert and Jonathan Brooks-Bartlett for feedback and support.  ... 
doi:10.1007/978-3-319-71078-5_23 fatcat:rioupqyebrflddwqo3rcg6oeqi

Using GANs for Sharing Networked Time Series Data: Challenges, Initial Promise, and Open Questions [article]

Zinan Lin, Alankar Jain, Chen Wang, Giulia Fanti, Vyas Sekar
2020 arXiv   pre-print
Limited data access is a longstanding barrier to data-driven research and development in the networked systems community.  ...  As a specific target, our focus in this paper is on time series datasets with metadata (e.g., packet loss rate measurements with corresponding ISPs).  ...  This work was supported in part by faculty research awards from Google and JP Morgan Chase, as well as a gift from Siemens AG.  ... 
arXiv:1909.13403v4 fatcat:lopcadvsn5ctjps4sujkdiw4tm

RDMAbox : Optimizing RDMA for Memory Intensive Workloads [article]

Juhyun Bae, Ling Liu, Yanzhao Wu, Gong Su, Arun Iyengar
2021 arXiv   pre-print
The optimizations are packaged in easy-to-use kernel and user space libraries for applications and systems in data center.  ...  We demonstrate the flexibility and effectiveness of RDMAbox by implementing a kernel remote paging system and a user space file system using RDMAbox. RDMAbox employs two optimization techniques.  ...  We implement a kernel remote paging system and a user-space network file system to show the performance of RDMAbox compared to existing best practices in both kernel and user space.  ... 
arXiv:2104.12197v2 fatcat:toc5mrykqnblzhevgwjj7pr5xi

A performance model of MVS

Willy W. Chiu-We, Min Chow
1978 IBM Systems Journal  
The current version of the MVS performance model allows two types of users, batch and TSO, and a number of system jobs, e.g., JES and TCAM/VTAM.  ...  8 TIME (SECONDS) TSO response density function for 30 TSO users case and 5 batch initia- tors case © MODELED 4 MEASURED 600-5 5 = 7 Ss TIME (SECONDS) storage.  ... 
doi:10.1147/sj.174.0444 fatcat:vux3skvmhve2tkxr3fh732kswe

Training Production Language Models without Memorizing User Data [article]

Swaroop Ramaswamy, Om Thakkar, Rajiv Mathews, Galen Andrew, H. Brendan McMahan, Françoise Beaufays
2020 arXiv   pre-print
Unlike all prior privacy-focused FL work of which we are aware, for the first time we demonstrate the deployment of a differentially private mechanism for the training of a production neural network in  ...  FL, as well as the instrumentation of the production training infrastructure to perform an end-to-end empirical measurement of unintended memorization.  ...  ACKNOWLEDGEMENTS The authors would like to specially thank Peter Kairouz, Ananda Theertha Suresh, Kunal Talwar, Abhradeep Thakurta, and our colleagues in Google Research for their helpful support of this  ... 
arXiv:2009.10031v1 fatcat:kdw2yg7eu5g4xdvtdlica52f2m

SRIFTY: Swift and Thrifty Distributed Neural Network Training on the Cloud

Liang Luo, Peter West, Pratyush Patel, Arvind Krishnamurthy, Luis Ceze
2022 Conference on Machine Learning and Systems  
Optimal VM selection that meets user constraints requires efficiently navigating a large search space while controlling for the performance variance associated with sharing cloud instances and networks  ...  Using insights from these studies, we built Srifty, a system that combines runtime profiling with learned performance models to accurately predict training performance and find the best VM choice that  ...  ACKNOWLEDGEMENTS This work was supported in part by the National Science Foundation under Grants #1518703 and #1723352, DARPA FastNICs, and gifts from Intel, VMWare, and Google.  ... 
dblp:conf/mlsys/LuoWPKC22 fatcat:d24f4gvxtfde3nwrlbxk4zgkgi

Dynamic Embeddings for Interaction Prediction [article]

Zekarias T. Kefato and Sarunas Girdzijauskas and Nasrullah Sheikh and Alberto Montresor
2021 arXiv   pre-print
In recommender systems (RSs), predicting the next item that a user interacts with is critical for user retention.  ...  Moreover, DeePRed's effectiveness comes from the aforementioned design and a multi-way attention mechanism that inspects user-item compatibility.  ...  Since our algorithm is designed in such a way that the short-term embeddings at time are not dependent on the ones at time − 1, batching is straightforward and DeePRed incurs in no overhead from batch  ... 
arXiv:2011.05208v2 fatcat:dlbsmewqhfgt5ce2fhntotf7fq

Interaction statistics from a database management system

J. D. Krinos
1973 Proceedings of the June 4-8, 1973, national computer conference and exposition on - AFIPS '73  
time-sharing systems have been in existence for several years.  ...  In data management technology there is a performance trade-off between query and updating which poses a design choice.  ... 
doi:10.1145/1499586.1499663 dblp:conf/afips/Krinos73 fatcat:xlakndc3hjcp7fim6wffrheshm

Srifty: Swift and Thrifty Distributed Training on the Cloud [article]

Liang Luo, Peter West, Arvind Krishnamurthy, Luis Ceze
2022 arXiv   pre-print
Optimal VM selection that meets user constraints requires efficiently navigating a large search space while controlling for the performance variance associated with sharing cloud instances and networks  ...  Using insights from these studies, we built Srifty, a system that combines runtime profiling with learned performance models to accurately predict training performance and find the best VM choice that  ...  ACKNOWLEDGEMENTS This work was supported in part by the National Science Foundation under Grants #1518703 and #1723352, DARPA FastNICs, and gifts from Intel, VMWare, and Google.  ... 
arXiv:2011.14243v3 fatcat:d5tkgfipgzbrnbegqwqxdf3hoa

Deep Learning Workload Scheduling in GPU Datacenters: Taxonomy, Challenges and Vision [article]

Wei Gao, Qinghao Hu, Zhisheng Ye, Peng Sun, Xiaolin Wang, Yingwei Luo, Tianwei Zhang, Yonggang Wen
2022 arXiv   pre-print
Deep learning (DL) shows its prosperity in a wide variety of fields. The development of a DL model is a time-consuming and resource-intensive procedure.  ...  However, traditional approaches designed for big data or high performance computing workloads can not support DL workloads to fully utilize the GPU resources.  ...  The model resource manager in TrIMS offers a multi-tiered cache for DL models to be shared across users' FaaS functions and enables DL model sharing across all levels of the memory hierarchy in the GPU  ... 
arXiv:2205.11913v3 fatcat:fnbinueyijb4nc75fpzd6hzjgq

Automated Fine-Grained CPU Provisioning for Virtual Machines

Davide B. Bartolini, Filippo Sironi, Donatella Sciuto, Marco D. Santambrogio
2014 ACM Transactions on Architecture and Code Optimization (TACO)  
This article tackles this issue with AutoPro: a runtime system that enhances IaaS clouds with automated and fine-grained resource provisioning based on performance SLOs.  ...  For this reason, users who deploy VMs subject to SLOs are usually prone to overprovisioning resources, thus resulting in inflated business costs.  ...  Anant Agarwal at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at the Massachusetts Institute of Technology (MIT), and was brought to Milano by Charles Gruenwald III via a winding  ... 
doi:10.1145/2637480 fatcat:bfdoikcfpvew7ktatofgimy7gm

Measuring the Importance of Collaborative Learning
IJARCCE - Computer and Communication Engineering

Deepika. S
2015 IJARCCE  
The problem of online multitask learning for solving multiple related classification tasks in parallel, aiming to classify every sequence of data received by each task accurately and efficiently.  ...  Second, classical classification methods, it is the batch or online, often encounter a dilemma when applied to a group of tasks i.e., a process of single classify model trained on the entire collection  ...  ACKNOWLEDGEMENT My sincere thanks to my guide Prof.S.Jenifer Asst.Professor Department of Computer Science and Engineering, Parisutham Institute of Technology and Science, Thanjavur for her help and guidance  ... 
doi:10.17148/ijarcce.2015.4225 fatcat:sk6s2g6h5ngqllq7fuvyz44yqy

Effects of Low-Quality Computation Time Estimates in Policed Schedulers

Justin M. Wozniak, Yingxin Jiang, Aaron Striegel
2007 40th Annual Simulation Symposium (ANSS'07)  
Researchers conducting computer simulations can often provide estimates of computation time for a given type of simulation, which may be used by the compute cluster to aid in resource allocation and scheduling  ...  In this paper, we examine models for simple job schedulers and examine the quality of the deadline guarantee given. Simulation results based on actual runtimes are provided and discussed.  ...  The actual computation time is obtained empirically by the grid; in this model, each task has a device that may be applied to an input set to determine computation time.  ... 
doi:10.1109/anss.2007.16 dblp:conf/anss/WozniakJS07 fatcat:zzepv3ymj5cozcdm5vox6vy4f4
« Previous Showing results 1 — 15 out of 29,468 results