Parallel Computing in Multiphase Flow Systems Simulations: Presented at 1994 International Mechanical Engineering Congress and Exposition, Chicago, Illinois, November 6-11, 1994
Jerry Bywaters, Lone Star Printmaker: A Study of His Print Notebook, With a Catalogue of His Prints and a Checklist of His Illustrations and Ephemeral Works
LA Revolucion Diabetica Del Dr. Atkins :El Innovador Programa para Prevenir y Controlar la Diabetes de Tipo 2 / Atkins Diabetes Revolution:The Groundbreaking Approach to Preventing And Controlling Typ
John Zahorjan - Seattle WA, US Derek L. Eager - Saskatoon, CA Mary K. Vernon - Madison WI, US
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
G06F015/16 H04N007/173
US Classification:
709231, 725 97, 725 87, 725 90
Abstract:
A method of efficiently transmitting streamed data of a program to multiple clients requesting the program at different times ranks the requests in a multilevel hierarchy, which describes merging of data streams servicing the requests. The multilevel hierarchy changes dynamically as new requests arrive or existing data streams are merged to reduce the bandwidth or other costs required to serve the requests. The hierarchy may be established by simple rules or by a modeling of the actual cost of possible hierarchies.
Method For Caching Of Media Files To Reduce Delivery Cost
Derek L. Eager - Saskatoon, CA Michael C. Ferris - Madison WI, US Mary K. Vernon - Madison WI, US
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
G06F015/16 H04N007/173
US Classification:
709231, 725 93
Abstract:
Improvement on delivery of continuous media programs, such as video, divides the programs into prefixes and suffixes of data. One of these groups, or other groups associated with alternate on-demand data streaming, may be stored at a remote location, while the other of the groups may be stored at a plurality of regional locations. Accordingly, when the media program is ordered by a plurality of clients, a portion of the program will be broadcast from the regional server, while the other portion of the program will be transmitted from the remote server. The prefixes and suffixes are chosen so as to maximize the efficiency of data transfer taking into account the cost of storing the data as well as bandwidth costs.
John Zahorjan - Seattle WA, US Derek L. Eager - Saskatoon, CA Mary K. Vernon - Madison WI, US
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
H04N 7/173
US Classification:
725 97
Abstract:
Preferred techniques for transmitting on-demand data files include transmitting a first data stream in response to a first client request, and a new data stream in response to a subsequent second client request. The second client may snoop increasing amounts data from the first data stream, thereby decreasing the second client's reliance on the second data stream until the client has merged with the first data stream, at which time transmission of the second data stream will be discontinued. Alternatively, a patch data stream may be broadcast to the second client having a bandwidth less than that of the first data stream. The second client, in addition to receiving the second data stream real-time, will record data from a patch data stream until the second client has merged with the first data stream, at which time transmission of the second data stream and patch data stream will be discontinued.
Method For Reduced Bandwidth For On-Demand Data Streaming Using Mini-Clusters
Derek L. Eager - Saskatoon, CA Mary K. Vernon - Madison WI, US
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
H04N 7/173
US Classification:
725101, 709231
Abstract:
An improvement on dynamic skyscraper delivery of continuous media programs, such as video, divides the channels used for the delivery of the video into leading and trailing groups. A cluster defining on transmission of a program can then be broken into mini-clusters in the leading group which may be freely matched to full clusters in the lower group with loosened alignment requirements. This decoupling provides more efficient allocation of bandwidth to on-demand consumer requests and permits strategic opportunities to merge requests with concurrently allocated bandwidth for similar programs.
Udi Manber - Tucson AZ Mary K. Vernon - Madison WI
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
G06F 1300 G06F 13362
US Classification:
395325
Abstract:
A first-come first-serve arbitration protocol arbitrates between processing agents seeking access to a common bus. The system is an extension of a parallel contention arbiter in which each agent has an identity value made up of two portions, one a fixed portion and one a variable portion. The variable portion of each agent's identity is incremented depending on how long the agent has been waiting for access to the bus and the variable portion is the most significant portion of each agent's identity. Thus in each arbitration in which the highest identity value is awarded control of the bus, the agent which has waited the longest will gain access to the bus prior to those which have not waited as long.
Round-Robin Protocol Method For Arbitrating Access To A Shared Bus Arbitration Providing Preference To Lower Priority Units After Bus Access By A Higher Priority Unit
Mary K. Vernon - Madison WI Udi Manber - Tuscon AZ
Assignee:
Wisconsin Alumni Research Foundation - Madison WI
International Classification:
G06F 1336
US Classification:
395725
Abstract:
A protocol is disclosed for the round-robin distributed arbitration of access to a common resource, such as a bus, in a multiprocessor system or machine. The protocol assigns identity values to each agent, or processor, which are compared during each arbitration to determine which agent is awarded control of the bus. To enhance fairness agents having an identity value lower than the winner of the last arbitration are favored over agents having a higher identity value in the next subsequent arbitration.