Person:
Ahamad, Mustaque

Permanent Link
https://hdl.handle.net/1853/71046
Associated Organization(s)
Organizational Unit
ORCID
ArchiveSpace Name Record

Filters

1993 - 1994 3
3
3
3
3
3
Reset filters

Settings
Author: John, Ranjit × End date: 2012 ×

Publication Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Evaluation of Causal Distributed Shared Memory for Data-race-free Programs
    (Georgia Institute of Technology, 1994) John, Ranjit ; Ahamad, Mustaque
    Distributed Shared Memory (DSM) is becoming an accepted abstraction for programming distributed systems. Although DSM simplifies the programming of distributed applications, maintaining a consistent view of shared memory operations across processors in a distributed system can be expensive. The causal consistency model of DSM can allow more efficient implementations of DSM because it requires that only causally ordered memory operations be viewed in the same order at different processors. Also, weakly ordered systems have been proposed which advocate the use of synchronization information to reduce the frequency of communication between processors. We have implemented a system that exploits both the weaker consistency of causal memory and the synchronization information used in weakly ordered systems. Consistency is ensured by locally invalidating data that is suspected to be causally overwritten and this is only done when certain synchronization operations complete at a processor. Data-race-free programs can be written in this system assuming that the system provided sequentially consistent memory. By implementing applications that have a variety of data sharing patterns, we show that performance comparable to message passing implementations of the applications can be achieved in the causal DSM system. The improved performance is due to a significant reduction in communication costs compared to the implementation of a strongly consistent memory system. These results show that causal memory can meet the consistency and performance requirements of many distributed applications.
  • Thumbnail Image
    Item
    Causal Memory: Implementation, Programming Support and Experiences
    (Georgia Institute of Technology, 1993) John, Ranjit ; Ahamad, Mustaque
    Distributed Shared memory (DSM) has become an accepted abstraction for programming distributed systems. Although DSM simplifies the programming of distributed applications, maintaining a consistent shared memory can be expensive. Weakly ordered systems which use synchronization information have been proposed to reduce the frequency of communication between processors. We have implemented a weakly ordered system based on the Causal memory mode. We provide language and runtime support which allow programs to run efficiently on Causal memory. Actual implementation results show a significant reduction in the number of messages when compared to a system maintaining a consistent shared memory.
  • Thumbnail Image
    Item
    An Evaluation of State Sharing Techniques in Distributed Operating Systems
    (Georgia Institute of Technology, 1993) John, Ranjit ; Ahamad, Mustaque ; Ramachandran, Umakishore ; Ananthanarayanan, R. (Rajagopal) ; Mohindra, Ajay
    A shared memory abstraction in distributed systems (DSM) provides ease of programming but could be costly to implement. Many protocols have been proposed recently that are based on different approaches for exploiting program semantics. We have implemented four different protocols that embody the different memory semantics and have evaluated them using applications that capture a wide range of state sharing patterns. Our main goal is to quantify the relative performance of these protocols with respect to representative applications. We expect the results of such a study to be useful for an operating system designer to choose the right protocol to support the DSM abstraction. One of the surprising results is that memory systems that provide weaker consistency or use synchronization information in coherence maintenance (e.g. release consistency) do not provide significantly better performance than the basic invalidation based protocol. The disparity between the performance using these protocols and programmer controlled state sharing can range from being insignificant to as high as 50% depending on application characteristics. The use of synchronization information in conjunction with user directives will bridge this performance gap but may decrease the programming ease of DSM.