MIMD References: EE686 Advanced Computer Architecture Design

Network Topology

An Overview of the NYU Ultracomputer Project (1986) (PDF)

The NYU Ultracomputer project was largely documented in a huge series of "Ultracomputer Reports" -- this is a circa 1986 collection of fragments from those. It's a pretty good reference for fetch-and-op.

@techreport{ gottlieb87overview,
    author = "Allan Gottlieb",
    title = "An Overview of the {NYU} Ultracomputer Project",
    number = "100",
    month = "1986 {"}Experimental Parallel Computing Architectures{"}, Editor - J.J. Dongarra. Elsevier Science Publishers; B.V. (North Holland)",
    year = "1987",
    url = "citeseer.ist.psu.edu/gottlieb86overview.html" }

A fetch-and-op implementation for parallel computers (PDF)

A nice paper with lots of circuit detail about fetch-and-op implementations.

@inproceedings{52443,
 author = {G. J. Lipovski and P. Vaughan},
 title = {A fetch-and-op implementation for parallel computers},
 booktitle = {ISCA '88: Proceedings of the 15th Annual International Symposium on Computer architecture},
 year = {1988},
 isbn = {0-8186-0861-1},
 pages = {384--392},
 location = {Honolulu, Hawaii, United States},
 publisher = {IEEE Computer Society Press},
 address = {Los Alamitos, CA, USA},
 }

Shared Memory Consistency Models: A Tutorial (PDF)

A nice overview of basic memory consistency issues.

@article{ adve96shared,
    author = "S. V. Adve and K. Gharachorloo",
    title = "Shared Memory Consistency Models: {A} Tutorial",
    journal = "IEEE Computer",
    volume = "29",
    number = "12",
    pages = "66--76",
    year = "1996",
    url = "citeseer.ist.psu.edu/adve95shared.html" }

TreadMarks: Distributed Shared Memory on Standard Workstations and Operating Systems (PS.gz, PDF)

One of the MANY papers about TreadMarks, which is probably the best known page-fault DSM system.

@inproceedings{treadmarks,
 author="P. Keleher and S. Dwarkadas and A.L. Cox and W. Zwaenepoel",
 title="TreadMarks: Distributed Shared Memory on Standard Workstations and Operating Systems",
 booktitle="Proceedings of the Winter 94 Usenix Conference",
 pages="115--131",
 month="January",
 year="1994"
}

Managing Polyatomic Coherence and Races with Replicated Shared Memory (PDF)

A little paper about an alternative implementation of what appears to be shared memory, but really is just remote memory access via function calls hidden by a few clever C++ templates.

@article{dietzrsm97,
 author="H. G. Dietz and T. I. Mattox",
 title="Managing Polyatomic Coherence and Races with Replicated Shared Memory"
 journal="Technical Committee on Computer Architecture (TCCA) Newsletter, Special Issue on Distributed Shared Memory and Related Issues",
 publisher="IEEE Computer Society",
 month="March",
 year="1997",
 url="http://tab.computer.org/tcca/NEWS/mar97/"
}