• Corpus ID: 13925247

Memory Block Relocation in Cache-Only Memory Multiprocessors

  title={Memory Block Relocation in Cache-Only Memory Multiprocessors},
  author={Gyungho Lee and Sujat Jamil},
  booktitle={Parallel and Distributed Computing and Systems},
COMA machine is similar to that in a traditional shared memory machine, there are a few aspects that dif ferentiate the AMs from the cache memory in traditional cache-coherent multiprocessors [8]. One important aspect unique to COMA is that the backing store of the AMs in a COMA machine is disks of secondary storage. So, unlike a traditional multiprocessor cache, write-back to the backing store at memory block replacement needs to be avoided. This creates a unique problem in a COMA, the needs… 

Figures from this paper

Performance characterization of shared attraction memories in cluster-based coma multiprocessors

It has been found that clustering significantly improves the performance of the attraction memory in a shared attraction memory machine and there is also a reduction in network traffic as well as the number of replacements.

A study of the efficiency of shared attraction memories in cluster-based COMA multiprocessors

From experiments with program-driven simulation it is found that clustering may improve the performance of the attraction memory significantly, and execution can remain efficient at higher memory pressure in clustered systems than in systems with single processor nodes.

Virtual memory on data diffusion architectures



An assessment of COMA multiprocessors

  • Gyungho Lee
  • Computer Science
    Proceedings of 9th International Parallel Processing Symposium
  • 1995
This paper studies the change in the miss ratio and the network traffic with the two coherence policies, update and invalidate, and shows that the two policies provide distinct characteristics which reveal different opportunity of improving COMA multiprocessors for different choice of coherence policy.

DDM - A Cache-Only Memory Architecture

The Data Diffusion Machine (DDM), a cache-only memory architecture (COMA) that relies on a hierarchical network structure, is described and simulated performance results are presented.

Evaluating the memory overhead required for COMA architectures

  • T. JoeJ. Hennessy
  • Computer Science
    Proceedings of 21 International Symposium on Computer Architecture
  • 1994
Simulation data shows that the frequency of data reshuffling is sensitive to the allocation policy and associativity of the memory but is relatively unaffected by the block size chosen, and that data replication in the attraction memory is important for good performance, but most gains can be achieved through replicated in the processor caches.

The Perfect Club Benchmarks: Effective Performance Evaluation of Supercomputers

A methodology for measuring the performance of supercomputers, including 13 Fortran programs that total over 50,000 lines of source code, and a set of guidelines that allow portability to several types of machines are presented.

DICE Prototype Specification”, DICE

  • Technical Report #6,
  • 1993

A Hypercube Shared Virtual Memory System

Unallocated Memory Space in COMA Multiprocessors

An important relation between the amount of unallocated space and the set associativity of the AM is identified, and the trade-off between additional unallocate memory space and higher set associative status is discussed.

DICE Prototy pe Specification

  • DICE Prototy pe Specification
  • 1993

A n Assessment of COMA Multiprocessor

  • Proceedings of the 9th International Parallel Processing Symposium,
  • 1995