Filters








6,490 Hits in 4.1 sec

Beyond Processor-centric Operating Systems

Paolo Faraboschi, Kimberly Keeton, Tim Marsland, Dejan S. Milojicic
2015 USENIX Workshop on Hot Topics in Operating Systems  
At rack scale, we can expect a large pool of non-volatile memory (NVM) that will be accessed by heterogeneous and decentralized compute resources [3, 17].  ...  In this paper, we describe the characteristics and consequences of memory-centric architectures and propose a memory-centric OS design that moves traditional OS functionality outside of the compute node  ...  Memory at large scale Addressing memory.  ... 
dblp:conf/hotos/FaraboschiKMM15 fatcat:wc4ehrbiffazzmdke2m6bmmma4

Composable architecture for rack scale big data computing

Chung-Sheng Li, Hubertus Franke, Colin Parris, Bulent Abali, Mukil Kesavan, Victor Chang
2017 Future generations computer systems  
A rack scale composable prototype system based on PCIe switch is described in Section 6. We describe the rack scale composable memory in Section 7.  ...  Figure 3 shows rack scale composability, which leverages the fast progress of the networking capabilities, software defined environments, and the increasing demand for high utilization of computing resources  ...  Pooling the memory in a centralized location, either at a rack scale or at datacenter scale, will impact memory performance and cost.  ... 
doi:10.1016/j.future.2016.07.014 fatcat:bf7rs7plxrb5beczlbljmze3dq

A Network Topology for Composable Infrastructures

Opeyemi O. Ajibola, Taisir E. H. El-Gorashi, Jaafar M. H. Elmirghani
2020 2020 22nd International Conference on Transparent Optical Networks (ICTON)  
This paper proposes a passive optical backplane as a new network topology for composable computing infrastructures.  ...  Resource disaggregation can be implemented physically or logically at different scales i.e. rack-scale, pod-scale or DC-scale [27] .  ...  consumption of computing capacity at the edge of the network.  ... 
doi:10.1109/icton51198.2020.9203275 fatcat:sdoscrg3zreytl4ob73vm4qytq

Web search for a planet: the google cluster architecture

L.A. Barroso, J. Dean, U. Holzle
2003 IEEE Micro  
Acknowledgments Over the years, many others have made contributions to Google's hardware architecture that are at least as significant as ours.  ...  At Google's scale, some limits of massive server parallelism do become apparent, such as the limited cooling capacity of commercial data centers and the less-than-optimal fit of current CPUs for throughput-oriented  ...  Thus, packing even more servers into a rack could be of limited practical use for large-scale deployment as long as such racks reside in standard data centers.  ... 
doi:10.1109/mm.2003.1196112 fatcat:g3n5yax2prap3pehcaq66ajflm

OS Support for Thread Migration and Distribution in the Fully Heterogeneous Datacenter

Pierre Olivier, Sang-Hoon Kim, Binoy Ravindran
2017 Proceedings of the 16th Workshop on Hot Topics in Operating Systems - HotOS '17  
image, a distributed shared memory system, and a heterogeneous scheduler.  ...  We propose a distributed OS architecture running on a fully heterogeneous computer cluster, enabling this cooperation through three main components: the abstraction of the entire cluster in a single system  ...  Based on these premises, we are reviving the case for the DSM in the rack-scale computing.  ... 
doi:10.1145/3102980.3103009 dblp:conf/hotos/OlivierKR17 fatcat:hhrwtk7tujhwtgxgz3zaae5bja

Domain Feature Mapping with YOLOv7 for Automated Edge-Based Pallet Racking Inspections

Muhammad Hussain, Hussain Al-Aqrabi, Muhammad Munawar, Richard Hill, Tariq Alsboui
2022 Sensors  
Pallet racking is an essential element within warehouses, distribution centers, and manufacturing facilities.  ...  Inspired by the trend toward smart industrial operations, we present a computer vision-based autonomous rack inspection framework centered around YOLOv7 architecture.  ...  Environmental Variance Modelling This architecture was aimed at pallet racking detection across a wide range of warehouses, distribution centers, and manufacturing facilities.  ... 
doi:10.3390/s22186927 pmid:36146273 fatcat:4z2ox5dxafeupaa3mooswllwc4

Network support for resource disaggregation in next-generation datacenters

Sangjin Han, Norbert Egi, Aurojit Panda, Sylvia Ratnasamy, Guangyu Shi, Scott Shenker
2013 Proceedings of the Twelfth ACM Workshop on Hot Topics in Networks - HotNets-XII  
This paper thus explores the question of whether we can build networks that enable disaggregation at datacenter scales.  ...  Datacenters have traditionally been architected as a collection of servers wherein each server aggregates a fixed amount of computing, memory, storage, and communication resources.  ...  It may be possible that mid-scale disaggregation (resource disaggregation at the rack-or pod-level) would provide enough efficiency and flexibility.  ... 
doi:10.1145/2535771.2535778 dblp:conf/hotnets/HanEPRSS13 fatcat:j7xuuwwp6jc4nmllbatpzog5oi

The Next Generation Of Exascale-Class Systems: The Exanest Project

R. Ammendolay, A. Biagioni, P. Cretaro, O. Frezza, F. Lo Cicero, A. Lonardo, M. Martinelli, P. S. Paolucci, E. Pastorelli, F. Simula, P. Vicini, G. Taffoni (+6 others)
2017 Zenodo  
Their goal is designing and implementing a physical rack prototype together with its cooling system, the storage non-volatile memory (NVM) architecture and a low-latency interconnect able to test different  ...  tested in large-scale ARM-based systems.  ...  This concept translates into the "affinity" that a given memory segment has with a particular computing core.  ... 
doi:10.5281/zenodo.823594 fatcat:td67qi6nibfl3ozhafhjk2ya7u

Optical Networks for Composable Data Centers [article]

Opeyemi O. Ajibola, Taisir E. H. El-Gorashi, Jaafar M. H. Elmirghani
2021 arXiv   pre-print
Composable data centers (DCs) have been proposed to enable greater efficiencies as the uptake of on-demand computing services grows.  ...  Relative to the implementation of a generic design that requires a (high capacity) dedicated transceiver on each point-to-point link on a mesh optical fabric in a composable DC rack, the targeted design  ...  Resource components in Rack 2 are physically disaggregated at rack-scale; the rack comprises of homogeneous compute nodes. Resource components in rack 3 are logically disaggregated at rack-scale.  ... 
arXiv:2106.04738v1 fatcat:ygbrukcmpjb2le22wcugwjvyiu

Configuration and Performance of a Beowulf Cluster for Large-Scale Scientific Simulations

M.K. Gobbert
2005 Computing in science & engineering (Print)  
At the University of Maryland, Baltimore County (UMBC), my colleagues and I bought a medium-sized 64processor cluster with high-performance interconnect and extended disk storage from IBM.  ...  implementation with the right hardware (in this case, the Beowulf cluster) can achieve parallel computing's two fundamental goals: to solve problems faster and to solve larger problems than we can on a serial computer  ...  This cluster constitutes a distributed-memory parallel computer-each node's memory can be accessed only by the CPUs on that node.  ... 
doi:10.1109/mcse.2005.29 fatcat:fpl46mov3bb3np4mns4mjospdi

X10 and APGAS at Petascale

Olivier Tardieu, Benjamin Herta, David Cunningham, David Grove, Prabhanjan Kambadur, Vijay Saraswat, Avraham Shinnar, Mikio Takeuchi, Mandana Vaziri
2014 SIGPLAN notices  
for RDMAs § congruent addresses required for RDMAs at scale § solution: dedicated memory allocator !  ...  Background § X10 tackles the challenge of programming at scale § HPC, cluster, cloud § scale out: run across many distributed nodes è this talk & PPAA talk § scale up: exploit multi-core and accelerators  ... 
doi:10.1145/2692916.2555245 fatcat:l57fpwlkyjciti4frjljsljhs4

X10 and APGAS at Petascale

Olivier Tardieu, Benjamin Herta, David Cunningham, David Grove, Prabhanjan Kambadur, Vijay Saraswat, Avraham Shinnar, Mikio Takeuchi, Mandana Vaziri
2014 Proceedings of the 19th ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '14  
for RDMAs § congruent addresses required for RDMAs at scale § solution: dedicated memory allocator !  ...  Background § X10 tackles the challenge of programming at scale § HPC, cluster, cloud § scale out: run across many distributed nodes è this talk & PPAA talk § scale up: exploit multi-core and accelerators  ... 
doi:10.1145/2555243.2555245 dblp:conf/ppopp/TardieuHCGKSSTV14 fatcat:kyw5snbvu5abjlk2eehtdx74ge

Power efficiency in high performance computing

Shoaib Kamil, John Shalf, Erich Strohmaier
2008 Proceedings, International Parallel and Distributed Processing Symposium (IPDPS)  
Our study provides power measurements for various computational loads on the largest scale HPC systems ever involved in such an assessment.  ...  This allows a less invasive approach for determining the power consumption of large-scale systems.  ...  The SGI ICE and Altix systems also use 48 VDC power distribution within the rack.  ... 
doi:10.1109/ipdps.2008.4536223 dblp:conf/ipps/KamilSS08 fatcat:x2tfqis7cffi5dhklt7rct46ju

Hoard: A Distributed Data Caching System to Accelerate Deep Learning Training on the Cloud [article]

Christian Pinto, Yiannis Gkoufas, Andrea Reale, Seetharami Seelam, Steven Eliuk
2018 arXiv   pre-print
Deep Learning system architects strive to design a balanced system where the computational accelerator -- FPGA, GPU, etc, is not starved for data.  ...  We describe the design and implementation of a distributed caching system called Hoard that stripes the data across fast local disks of multiple GPU nodes using a distributed file system that efficiently  ...  The distributed cache layer at the bottom realizes the distributed cache on data-center nodes.  ... 
arXiv:1812.00669v1 fatcat:rfdhznhoyfbcndkwpp7efnptpy

The TianHe-1A Supercomputer: Its Hardware and Software

Xue-Jun Yang, Xiang-Ke Liao, Kai Lu, Qing-Feng Hu, Jun-Qiang Song, Jin-Shu Su
2011 Journal of Computer Science and Technology  
TH-1A is now deployed in National Supercomputer Center in Tianjin and provides high performance computing services.  ...  Aiming at the problems of large scale heterogeneous parallel programming, such as program segmentation, data distribution, processes synchronization, load balancing, performance optimization, TH-HPI uses  ...  There are 140 racks in the whole TH-1A system, including 112 compute racks, 8 service racks, 6 communication racks, and 14 I/O racks. The entire system occupies 700 m 2 .  ... 
doi:10.1007/s02011-011-1137-8 fatcat:jd3fxx7qr5bllfmhrafbq4ftb4
« Previous Showing results 1 — 15 out of 6,490 results