Download:
|
by S. Konstantinidou, Grace Ngai
http://www.cs.jhu.edu/~gyn/publications/flow-control/flow-control.ps
Add To MetaCart
Abstract:
In network-based parallel architectures, the issues of fairness, freedom of deadlock due to finite buffers and guaranteed message delivery can affect not only performance but even more importantly, the guarantees that the architecture provides to the users so that they can write correct, race-free portable programs. When blocking communication is assumed, the solutions to these problems can be relatively simple and efficient. When nonblocking communications must be supported for performance, these issues complicate the designs of architectures irrespective of the programming paradigm that the architecture supports and the characteristics of the network. In this paper, we consider blocking and nonblocking communication and the effect of flow control in the latter case, on both performance and correctness. Our experimental results are derived on a cluster of dual processor SMPs connected via FDDI fiber ring. The cluster is built on off-the-shelf hardware components, although communication protocols customized for parallelism, and in particular the remote read/write programmingparadigm, have been developed and are executed by one of the processors while the other is dedicated to computation.
Citations
|
318
|
The Stanford FLASH Multiprocessor
– Kuskin, Ofelt, et al.
- 1994
|
|
269
|
Virtual memory mapped network interface for the SHRIMP multicomputer
– Blumrich, Li, et al.
- 1994
|
|
207
|
Weak Ordering -- A New Def-inition
– Adve, Hill
- 1990
|
|
154
|
Parallel programming in split-c
– Culler, Dusseau, et al.
- 1993
|
|
99
|
Integrating message-passing and shared-memory: Early experience
– Kranz, Johnson, et al.
- 1993
|
|
94
|
Synchronization, Coherence, and Event Ordering in Multiprocessors
– Dubois, Scheurich, et al.
- 1988
|
|
66
|
Directory-based cache coherence in large-scale multiprocessors
– Chaiken, Fields, et al.
- 1990
|
|
61
|
Where is Time Spent in Message-Passing and Shared-Memory Programs
– Chandra, Larus, et al.
- 1994
|
|
58
|
Integration of message passing and shared memory in the Stanford FLASH multiprocessor
– Heinlein, Gharachorloo, et al.
- 1994
|
|
42
|
A case for NOW
– Anderson, Culler, et al.
- 1995
|
|
40
|
Deadlock avoidance in store-and-forward networks. 1: Store-and-forward deadlock
– Merlin, Schweitzer
- 1980
|
|
26
|
The SP1 High Performance Switch
– Stunkel, Shea, et al.
- 1994
|
|
26
|
Multiprocessor System Architectures
– Catanzaro
- 1994
|
|
23
|
A Comparison of Message Passing and Shared Memory Architectures for Data Parallel Programs
– Klaiber, Levy
- 1994
|
|
17
|
Low-Latency Message Communication Support for the AP1000
– Shimizu, Horie, et al.
- 1992
|
|
12
|
Bounds on the efficiency of message-passing protocols for parallel computers
– Cypher, Konstantinidou
- 1993
|
|
12
|
AP1000+: Architectural Support of Put/Get Interface for Parallelizing Compiler
– Hayashi, Doi, et al.
- 1994
|
|
4
|
Software vs. Hardware Shared Memory Implementation: A Case Study
– Cox, Dwarkadas, et al.
- 1994
|
|
4
|
Delta message passing protocol
– Regnier
- 1992
|
|
2
|
The Network
– Pierre, Wong, et al.
- 1992
|
|
1
|
The CRAY T3D Massively Parallel Processing System", available on the www at URL http://www.cray.com/PUBLIC/product-info/mpp
– Inc
|
|
1
|
Research Incorporated. "SHMEM Technical Note for C." Revision 2.3
– Cray
|
