2.
    发明专利
    未知

    公开(公告)号:DE69219848D1

    公开(公告)日:1997-06-26

    申请号:DE69219848

    申请日:1992-08-10

    Applicant: IBM

    Abstract: Methods and apparatus are set forth which optimize and balance the use of processor card (or complex) resources (e.g., memory, the local bus, etc.), with the use of other system resources, such as the card to card communications bus and devices attached thereto, in dual bus computing systems. A new data rate management technique reduces data transfer overhead particularly for burst mode transfers. The invention promptly services pending memory refresh requests; limits multiple accesses to on card (or processor complex) memory by an Alternate Bus Master to a predetermined number of cycles where the processor requests the use of its local bus; and allows an Alternate Bus Master unlimited accesses to the processor local bus when the Alternate Bus Master owns the card to card communications bus and the processor subsequently requests that bus. The aforementioned balance of resources is achieved using a Bus Hold On Grant ("BSHOG") scheduling procedure which, assuming a bursting Alternate Bus Master has gained control of the local processor bus and is conducting data transfer cycles when the processor requests the local bus, allows the Alternate Bus Master to conditionally retain local bus ownership over a plurality of Alternate Bus Master data transfer cycles.

    ALTERNATE MASTER BURSTING DATA RATE MANAGEMENT TECHNIQUES FOR USE IN COMPUTER SYSTEMS HAVING DUAL BUS ARCHITECTURE

    公开(公告)号:CA2068010C

    公开(公告)日:1996-10-22

    申请号:CA2068010

    申请日:1992-05-05

    Applicant: IBM

    Abstract: Methods and apparatus are set forth which optimize and balance the use of processor card (or complex) resources (e. g., memory, the local bus, etc .), with the use of other system resources, such as the card to card communications bus and devices attached thereto, in dual bus computing systems. A new data rate management technique reduces data transfer overhead particularly for burst mode transfers. The invention promptly services pending memory refresh requests; limits multiple accesses to on card (or processor complex) memory by an Alternate Bus Master to a predetermined number of cycles where the processor requests the use of its local bus; and allows an Alternate Bus Master unlimited accesses to the processor local bus when the Alternate Bus Master owns the card to card communications bus and the processor subsequently requests that bus. The aforementioned balance of resources is achieved using a Bus Hold On Grant ("BSHOG") scheduling procedure which, assuming a bursting Alternate Bus Master has gained control of the local processor bus and is conducting data transfer cycles when the processor requests the local bus, allows the Alternate Bus Master to conditionally retain local bus ownership over a plurality of Alternate Bus Master data transfer cycles.

    4.
    发明专利
    未知

    公开(公告)号:DE69219848T2

    公开(公告)日:1997-10-23

    申请号:DE69219848

    申请日:1992-08-10

    Applicant: IBM

    Abstract: Methods and apparatus are set forth which optimize and balance the use of processor card (or complex) resources (e.g., memory, the local bus, etc.), with the use of other system resources, such as the card to card communications bus and devices attached thereto, in dual bus computing systems. A new data rate management technique reduces data transfer overhead particularly for burst mode transfers. The invention promptly services pending memory refresh requests; limits multiple accesses to on card (or processor complex) memory by an Alternate Bus Master to a predetermined number of cycles where the processor requests the use of its local bus; and allows an Alternate Bus Master unlimited accesses to the processor local bus when the Alternate Bus Master owns the card to card communications bus and the processor subsequently requests that bus. The aforementioned balance of resources is achieved using a Bus Hold On Grant ("BSHOG") scheduling procedure which, assuming a bursting Alternate Bus Master has gained control of the local processor bus and is conducting data transfer cycles when the processor requests the local bus, allows the Alternate Bus Master to conditionally retain local bus ownership over a plurality of Alternate Bus Master data transfer cycles.

    5.
    发明专利
    未知

    公开(公告)号:DE69836307T2

    公开(公告)日:2007-04-26

    申请号:DE69836307

    申请日:1998-09-15

    Applicant: IBM

    Abstract: A pair of communications adapters each include a number of digital signal processors (12) and network interface circuits (24) for the attachment of a multi-channel telephone line. A bus connecting the communications adapters can carry data between a network line attached to one of the adapters and the digital signal processors of the other adapter. The digital signal processors on each card are connected to a host, or controller, processor (2). Each digital signal processor interrupts its host processor by transmitting an interrupt control block as data to a data memory of the host processor, and by subsequently sending an interrupt causing the host processor to examine the data memory. Preferably, the interrupt control block includes data representing a number of requested interrupts.

    6.
    发明专利
    未知

    公开(公告)号:DE69836307D1

    公开(公告)日:2006-12-14

    申请号:DE69836307

    申请日:1998-09-15

    Applicant: IBM

    Abstract: A pair of communications adapters each include a number of digital signal processors (12) and network interface circuits (24) for the attachment of a multi-channel telephone line. A bus connecting the communications adapters can carry data between a network line attached to one of the adapters and the digital signal processors of the other adapter. The digital signal processors on each card are connected to a host, or controller, processor (2). Each digital signal processor interrupts its host processor by transmitting an interrupt control block as data to a data memory of the host processor, and by subsequently sending an interrupt causing the host processor to examine the data memory. Preferably, the interrupt control block includes data representing a number of requested interrupts.

Patent Agency Ranking