Filters








4,076 Hits in 3.4 sec

Leveraging torus topology with deadlock recovery for cost-efficient on-chip network

Minjeong Shin, John Kim
2011 2011 IEEE 29th International Conference on Computer Design (ICCD)  
As a result, deadlocks can be exactly detected without having to rely on a timeout mechanism and when needed, recover from the deadlock.  ...  In this work, we propose deadlock recovery with tokens (DRT) in on-chip networks that exploits on-chip networks -exploiting the abundant wires available while minimizing the need for additional buffers  ...  Both parts uses a separate network to reduce the impact of deadlock recovery on the design of the baseline, data network. A.  ... 
doi:10.1109/iccd.2011.6081371 dblp:conf/iccd/ShinK11 fatcat:cxqpzgzslzh67eesiwyifxgq4a

Exploring Fault-Tolerant Network-on-Chip Architectures

Dongkook Park, C. Nicopoulos, Jongman Kim, N. Vijaykrishnan, C.R. Das
2006 International Conference on Dependable Systems and Networks (DSN'06)  
In this paper, we examine the impact of transient failures on the reliability of on-chip interconnects and develop comprehensive counter-measures to either prevent or recover from them.  ...  The advent of deep sub-micron technology has exacerbated reliability issues in on-chip interconnects.  ...  specified in Equation (on buffer (at node ) Total size of the Retransmission buffer (at node ) Number of nodes involved in the deadlock Number of flits per packet (When a deadlock is detected, the transmission  ... 
doi:10.1109/dsn.2006.35 dblp:conf/dsn/ParkNKVD06 fatcat:v3gyhdseqzev5oglouxl62ta6m

Parallel Deadlock Detection and Recovery for Networks-on-Chip Dedicated to Diffused Computations

Pierre Bomel, Marc Sevaux
2013 2013 Euromicro Conference on Digital System Design  
This article concerns the first step towards the future design of routing strategies: the proposal of a deadlock detection and recovery algorithm.  ...  FPGAs allow the creation of many-cores architectures and, because they are reconfigurable, can embed networks-on-a-chip (NoCs) that have been finely tuned for these kernel's specificities.  ...  The cost of the deadlock escape channel is in O(S.V.N) because each node (one of N) stores one packet (size S) per VC (one of V).  ... 
doi:10.1109/dsd.2013.14 dblp:conf/dsd/BomelS13 fatcat:kzowpqqi4zbo3brmag727dowqe

Distributed resolution of network congestion and potential deadlock using reservation-based scheduling

Yong Ho Song, T.M. Pinkston
2005 IEEE Transactions on Parallel and Distributed Systems  
)-has a substantial impact on overall system performance.  ...  They should have an efficient mechanism to detect the existence of deadlocks in order to avoid unnecessary performance degradation [17, 16] .  ...  The figure also shows the impact of deadlock handling techniques on the performance of networks using a previously proposed injection limitation technique, called ALO (At Least One) [3] .  ... 
doi:10.1109/tpds.2005.93 fatcat:oojowp6qjffc3fx6yaqsbo2bp4

Deadlock-Recovery Support for Fault-tolerant Routing Algorithms in 3D-NoC Architectures

Akram Ben Ahmed, Achraf Ben Ahmed, Abderazek Ben Abdallah
2013 2013 IEEE 7th International Symposium on Embedded Multicore Socs  
RAB detects the presence of deadlock in the buffer and removes it with no considerable performance drop.  ...  The proposed deadlock-recovery technique was implemented on our earlier designed 3D-NoC system (3D-OASIS-NoC 1 ), which adopts Look-Ahead-Fault-Tolerant routing algorithm (LAFT).  ...  In each input-port, a buffer-controller (BC) manages the detection of deadlock and handles the assignment of wr-adr and rdadr addresses.  ... 
doi:10.1109/mcsoc.2013.39 fatcat:4hzstec7wzcxxcfqxgkry6o4kq

Deadlock-avoidance for streaming applications with split-join structure: Two case studies

Peng Li, Kunal Agrawal, Jeremy Buhler, Roger D. Chamberlain, Joseph M. Lancaster
2010 ASAP 2010 - 21st IEEE International Conference on Application-specific Systems, Architectures and Processors  
When these computations are mapped onto real parallel platforms, however, some computations, especially ones in which some nodes act as filters, can deadlock the system due to finite buffering on channels  ...  For systems with large buffers or a low filtering ratio, the communication overhead of the Non-propagating Algorithm is negligible.  ...  [q] is computed as the minimum total buffer size of other paths divided by the number of filtering nodes on this path, as shown in Algorithm 2.  ... 
doi:10.1109/asap.2010.5540957 dblp:conf/asap/LiABCL10 fatcat:xgfp6mnvn5fxxiu76eimkhym4m

Deadlock-free buffer configuration for stream computing

Peng Li, Jonathan C Beard, Jeremy D Buhler
2016 The international journal of high performance computing applications  
However, the connection between the configuration of output buffers and application deadlocks has not been studied.  ...  We also propose an efficient method based on all-pair shortest path algorithms to detect unsafe buffer configurations. We also sketch a method to adjust an unsafe buffer configuration to a safe one.  ...  Let |q|o be the size of the output buffer for channel q, and set |q| ′ o = |q|o − 1. Let |q| be the total buffer size of q.  ... 
doi:10.1177/1094342016675679 fatcat:w2ht4rzksvft7gmv3sym4vzjse

FlexVC: Flexible Virtual Channel Management in Low-Diameter Networks

Pablo Fuentes, Enrique Vallejo, Ramon Beivide, Cyriel Minkenberg, Mateo Valero
2017 2017 IEEE International Parallel and Distributed Processing Symposium (IPDPS)  
This restricts the number of buffer resources depending on the routing mechanism and limits performance due to an inefficient use.  ...  Deadlock avoidance mechanisms for lossless lowdistance networks typically increase the order of virtual channel (VC) index with each hop.  ...  ACKNOWLEDGMENT This work has been supported by the Spanish Government  ... 
doi:10.1109/ipdps.2017.110 dblp:conf/ipps/Fuentes0BMV17 fatcat:37jse6f2efabzbiev3prudcqpq

Deadlock-free buffer configuration for stream computing

Peng Li, Jonathan Beard, Jeremy Buhler
2015 Proceedings of the Sixth International Workshop on Programming Models and Applications for Multicores and Manycores - PMAM '15  
However, the connection between the configuration of output buffers and application deadlocks has not been studied.  ...  We also propose an efficient method based on all-pair shortest path algorithms to detect unsafe buffer configurations. We also sketch a method to adjust an unsafe buffer configuration to a safe one.  ...  Let |q|o be the size of the output buffer for channel q, and set |q| ′ o = |q|o − 1. Let |q| be the total buffer size of q.  ... 
doi:10.1145/2712386.2712403 dblp:conf/ppopp/LiBB15 fatcat:tghn2rid65hrlndm3ba7fei5zy

A progressive approach to handling message-dependent deadlock in parallel computer systems

Yong Ho Song, T.M. Pinkston
2003 IEEE Transactions on Parallel and Distributed Systems  
One of the crucial requirements to achieve high network performance is to efficiently handle deadlock anomalies.  ...  The proposed technique relaxes restrictions considerably, allowing the routing of packets and the handling of message-dependent deadlocks to be much more efficient-particularly when network resources are  ...  The authors also gratefully acknowledge the helpful suggestions made by José Duato and the anonymous reviewers.  ... 
doi:10.1109/tpds.2003.1189584 fatcat:ptwt32tflngsllwfpifhhhcrci

A deadlock detection mechanism for true fully adaptive routing in regular wormhole networks

Soojung Lee
2007 Computer Communications  
In order to alleviate the overhead of running a recovery procedure, the studies on deadlock detection have focused on the accuracy of deadlock detection, trying to reduce the number of false detections  ...  The proposed detection scheme is based on the turn model and designed to declare only one packet per simple cycle of blocked packets as deadlocked.  ...  The central deadlock buffers for Disha Sequential have the size of two flits.  ... 
doi:10.1016/j.comcom.2007.02.013 fatcat:6uayn4d3pnfc7exnohi4lv5j6e

Worm-Bubble Flow Control

Lizhong Chen, T. M. Pinkston
2013 2013 IEEE 19th International Symposium on High Performance Computer Architecture (HPCA)  
Wormhole switching allows the channel buffer size to be smaller than the packet size, thus is preferred by on-chip networks.  ...  In this paper, we propose Worm-Bubble Flow Control (WBFC), a new flow control scheme that can avoid deadlock in wormholeswitched tori using minimally 1-flit-sized buffers per VC and one VC in total.  ...  Acknowledgements We sincerely thank the anonymous reviewers for their helpful comments and suggestions.  ... 
doi:10.1109/hpca.2013.6522333 dblp:conf/hpca/ChenP13 fatcat:zotvjad3kfhyvpkl7ulcxd7dze

Precise Dynamic Analysis for Slack Elasticity: Adding Buffering without Adding Bugs [chapter]

Sarvani Vakkalanka, Anh Vo, Ganesh Gopalakrishnan, Robert M. Kirby
2010 Lecture Notes in Computer Science  
Increasing the amount of buffering for MPI sends (increasing the eager limit) is an effective way to improve the performance of MPI programs.  ...  Previous work did not provide any characterization of the space of slack elastic programs -those programs for which buffering can be safely added.  ...  . • The new version of ISP containing POE MSE is shown to perform efficiently on non-trivially sized MPI programs.  ... 
doi:10.1007/978-3-642-15646-5_16 fatcat:4ye2zipzpnhxxm4vebqfzjhvr4

Compressionless routing: a framework for adaptive and fault-tolerant routing

J.H. Kim, Ziqiang Liu, A.A. Chien
1997 IEEE Transactions on Parallel and Distributed Systems  
to a wide variety of network topologies, and elimination of the need for buffer allocation messages.  ...  CR exploits the tight-coupling between wormhole routers for flow control to detect and recover from potential deadlock situations.  ...  ACKNOWLEDGMENTS We would like to thank the reviewers for their valuable comments and suggestions. We also thank the members of the Concurrent Systems Architecture Group for helpful discussions.  ... 
doi:10.1109/71.584089 fatcat:ky5xsjitwndo5g7v4fas6zmnom

Compressionless routing

J. H. Kim, Z. Liu, A. A. Chien
1994 SIGARCH Computer Architecture News  
to a wide variety of network topologies, and elimination of the need for buffer allocation messages.  ...  CR exploits the tight-coupling between wormhole routers for flow control to detect and recover from potential deadlock situations.  ...  ACKNOWLEDGMENTS We would like to thank the reviewers for their valuable comments and suggestions. We also thank the members of the Concurrent Systems Architecture Group for helpful discussions.  ... 
doi:10.1145/192007.192050 fatcat:44ndfht5uzf2zecxkfzip425ha
« Previous Showing results 1 — 15 out of 4,076 results