摘要:
A technique for negotiating the width of a link between a first device and a second device includes detecting, during initialization, a respective signal on one or more control lines associated with at least a portion of an N-bit link. The N-bit link is configured as a single link having a width of N or multiple sublinks having a width less than N based on a respective value of the respective signal on the one or more control lines.
摘要:
A peripheral interface circuit for an I/O node of a computer system. A peripheral interface circuit for an input/output node of a computer system includes a first buffer circuit, a second buffer circuit and a bus interface circuit. The first buffer circuit receives packet commands and may include a first plurality of buffers each corresponding to a respective virtual channel of a plurality of virtual channels. The second buffer circuit is coupled to receive packet commands from the bus interface circuit and may include a second plurality of buffers each corresponding to a respective virtual channel of the plurality of virtual channels. The bus interface circuit may be configured to translate selected packet commands stored in the first buffer circuit into commands suitable for transmission on a peripheral bus.
摘要:
A data bus is divided into two portions. One portion of the bus transfers data from one side of the bus to the other and the other portion of the bus transfers data in the opposite direction. Bus cycles that originate from one side of the bus only go in one direction (from the originator to the other side). In order to avoid inefficiency because one of the portions of the bus may become unused if a long bus cycle is going in one direction while nothing is being transferred in the opposite direction, one side can take over the whole data bus and transfer data over both sides of the bus.
摘要:
A system and method for observing transactions on a packet bus is disclosed. In one embodiment, a computer system includes a plurality of input/output (I/O) nodes serially coupled to a processor. Each of the I/O nodes may be configured to operate in a first (normal) mode, and a second (analysis) mode. During the normal mode, packets may be selectively conveyed through an I/O tunnel in the I/O node, and particular packets may be selectively conveyed to a peripheral bus interface in the I/O node. In the analysis mode, electrical signals corresponding to packets conveyed through the I/O tunnel may be replicated on a peripheral bus coupled to the peripheral bus interface. No conversion from the packet bus protocol to the peripheral bus protocol. A signal analyzer may be coupled to the peripheral bus, thereby allowing observation of the electrical signals.
摘要:
A computer system I/O node. An input/output node for a computer system includes a first receiver unit configured to receive a first command on a first communication path and a first transmitter unit coupled to transmit a first corresponding command that corresponds to the first command on a second communication path. The input/output node also includes a second receiver unit configured to receive a second command on a third communication path and a second transmitter unit coupled to transmit a second corresponding command that corresponds to the second command on a fourth communication path. Further, the input/output node includes a bridge unit coupled to receive selected commands from the first receiver and the second receiver and configured to transmit commands corresponding to the selected commands upon a peripheral bus.
摘要:
A system and method are presented for equalizing data buffer storage and fetch rates of peripheral devices. A computer system of the present invention includes a central processing unit (CPU), first and second peripheral devices, and a data buffer. The first peripheral device stores data within the data buffer, and the second peripheral device fetches data from the data buffer. A fraction of the data buffer contains unread data (i.e. data stored within the data buffer by the first peripheral device and not yet fetched by the second peripheral device). The first peripheral device includes a reload register, the contents of which determines the rate at which the first peripheral device stores data within the data buffer. The CPU produces a reload value, which is stored within the reload register, such that the rate at which the first peripheral device stores the data within the data buffer is made substantially equal to the rate at which the second peripheral device fetches the data from the data buffer. The data buffer is preferably operated a first-in-first-out manner, and includes a write pointer and a read pointer. The CPU preferably produces the reload value such that approximately half the memory locations within the data buffer contain unread data at any given time.
摘要:
A computer system is presented having various peripheral devices coupled to a PCI local bus (i.e., an expansion bus), a subset of the peripheral devices having quaternary interfaces configured to communicate via quaternary signals conveyed upon the PCI local bus. The various peripheral devices may include a video/graphics card, a sound card, a hard disk drive, a CD-ROM drive, and a network interface card. When data is transferred from a master device to a target device, and the master and target devices both have quaternary interfaces, the master device converts the data to quaternary signals before transmitting the data the target device via the PCI local bus. The target device receives the quaternary signals from the PCI local bus and converts the quaternary signals to the binary data. Two binary digits (i.e., bits) of information are advantageously conveyed using quaternary signals in the time required to transmit a single bit using binary signals, thus providing increased efficiency and reduced bus bandwidth requirements. If either device does not include a quaternary interface, the data is transferred using conventional binary signals. A handshaking protocol may be used to determine if both the master and target devices include a quaternary interface. The handshaking protocol is implemented using handshaking signals conveyed over additional control lines added to the PCI local bus. Alternatively, a configuration memory may be included in the quaternary interface of the master device to reduce the required number of additional control lines from two to one.
摘要:
A triggering circuit obviates propagation differences in differential combinatorial logic clocking of upstream and downstream state machines ("SM's"). The triggering circuit imposes an output state on the downstream SM in response to the appearance of an appropriate combination of the upstream SM output state and selected data at the triggering circuit input. The triggering circuit and the upstream SM are clocked from a common signal preventing the upstream SM from changing state before the triggering circuit produces the proper signal to impose the expected state on the downstream SM. The downstream SM takes on the correct output state in dependable correspondence with a selected upstream SM output state. In a preferred embodiment, a D flip-flop generates a triggering signal in response to a selected combination of upstream SM output state and system data. The D flip-flop triggering signal imposes a selected output state on a downstream SM through the asynchronous SET and CLEAR inputs of the downstream SM flip-flops. Because the D flip-flop and upstream SM are both clocked off the same trailing edge of the WRITE line, the upstream SM and D flip-flop change state together, preventing the upstream SM from changing state before the triggering signal is generated.
摘要:
A computer system includes a first and a second integrated circuit coupled by a communication link. The communication link operates in a power savings mode in which data is not transmitted over the link. Periodically, the communication link enters a training phase in which training patterns are transmitted over the communication link for a predetermined time period. The communication link returns to the power savings mode after the predetermined time period has elapsed. At least one sideband signal, separate from the communication link, and coupled between the first and second integrated circuits, is used to indicate when to enter the training phase from the power savings mode and exit the training phase and return to the power savings mode.
摘要:
A method and apparatus for guaranteed data synchronization. In one embodiment, a data synchronization unit includes a memory unit, a write pointer unit, a read pointer unit, and synchronization pulse logic. The memory unit may receive information from a source external to the data synchronization unit. The write pointer may specify an address within the memory where incoming information is to be written. A read pointer outgoing information is to be read from. The data synchronization unit may also include synchronization pulse logic. The synchronization pulse logic may be configured to, in a synchronization mode, to generate a synchronization pulse. In response to an assertion of the synchronization pulse, the read pointer may be initialized such that data read out of the memory unit is valid (i.e. guaranteed to meet electrical requirements) at that time.