Rabindra P Kar

US Patent:

57614393, Jun 2, 1998

Filed:

Sep 25, 1995

Appl. No.:

8/533564

Inventors:

Rabindra P. Kar - Beaverton OR
Kenneth W. Reese - Portland OR

Assignee:

Intel Corporation - Santa Clara CA

International Classification:

G06F 112

US Classification:

39520078

Abstract:

In a computer network including a first computer communicatively coupled with at least a second computer through data and control channels, wherein each of the first and second computers are provided with an application program, a communication program and a synchronization program. In each of the first and second computers, the communication program is provided between the application program and the data and control channels, while the synchronization program is provided between the application program and the communication program. The synchronization program including means for sequentially forwarding commands and data units received from the application program of the same computer to the communication program of the same computer for transmission to another computer as corresponding control and data signals on the control and data channels, respectively. The synchronization program further comprising means for synchronizing the control signals and data signals sent on the control and data channels, respectively, by deferring sequential forwarding of a next sequential command to the communication program until an acknowledgment is received from the other computer acknowledging receipt of all previously transmitted control and data signals.

US Patent:

60120827, Jan 4, 2000

Filed:

Sep 23, 1997

Appl. No.:

8/935742

Inventors:

Rabindra P. Kar - Beaverton OR

Assignee:

Intel Corporation - Santa Clara CA

International Classification:

G06F 900

US Classification:

709107

Abstract:

A method and apparatus for CPU cycle stealing on a non-preemptive multi-tasking operating system allowing a first application to preempt other applications which are running concurrently on the operating system. The steps taken to perform the CPU cycle stealing include: inserting a hook near a starting address of message monitoring calls, and vectoring the execution path to the first application whenever one of the other applications calls the message monitoring calls, allowing the first application to steal CPU cycles from one of the other applications. The vectoring is provided by the hook inserted near the starting address of the message monitoring calls.