摘要:
A method and system to adapt communication links statically and/or dynamically to their individual link conditions on a platform. The communicatively coupled devices have logic to adapt one or more settings of a respective one or more communication links with another device based at least in part on a respective metric of received data patterns from the respective one or more communication links. The communicatively coupled devices in the platform have a back channel to allow feedback or information to be sent from one receiving device to a transmitting device in one embodiment of the invention.
摘要:
An optical memory extension architecture. A first electrical logic circuit on a first die communicates data according to a packetized, point-to-point interconnect protocol at a full data rate. A first gasket circuit is coupled to receive the data from the first electrical logic circuit. The first gasket circuit causes the data to be converted to an optical format to be transmitted at a rate that is at least double the full data rate. A second gasket circuit is coupled to receive the data in the optical format from the first gasket circuit. The second gasket circuit causes the data to be converted to an electrical format conforming to the packetized, point-to-point interconnect protocol. A second electrical logic circuit on a second die is coupled to receive the data from the first electrical logic circuit through the first gasket circuit and the second gasket circuit.
摘要:
Methods and apparatus relating to link power savings with state retention are described. In one embodiment, one or more components of two agents coupled via a serial link are turned off during idle periods while retaining link state in each agent. Other embodiments are also disclosed.
摘要:
A periodic control window is embedded in a link layer data stream to be sent over a serial data link, where the control window is configured to provide physical layer information including information for use in initiating state transitions on the data link. The link layer data can be sent during a link transmitting state of the data link and the control window can interrupt the sending of flits. In one aspect, the information includes link width transition data indicating an attempt to change the number of active lanes on the link.
摘要:
A method and system to reduce the power supply noise of a platform during the training of high speed communication links. In one embodiment of the invention, the device has logic to stagger a bit lock pattern for each of one or more communication links and scramble a training sequence for each of the one or more communication links. By doing so, it removes the need for anti-noise circuits and in turn, reduces the silicon area and power of the devices. Further, by having the logic in the physical layers to facilitate the training of the communication links, it eliminates the need to redesign the package of the devices to shift the resonant frequencies.
摘要:
Briefly, in accordance with one embodiment of the invention, a method of performing high speed signaling includes the following. A preamble signal and an end of packet (EOP) signal are transmitted at a low frequency using rail-to-rail voltage signal levels. Later, high frequency signaling is transmitted using a voltage signal level swing that is less than rail-to-rail.
摘要:
A RAM buffer controller for managing the address input lines of a RAM buffer to simulate the operation of two FIFO's therein. Apparatus is included for allowing random access by a node processor in a local area network node using the RAM buffer controller to manage "transmit and receive" FIFO's to have random access to any address in the address space of the buffer without restriction to FIFO boundaries. Apparatus is also included for transmitting packets from said buffer organized into one or two linked lists. Further, apparatus is included for allowing independent initialization of any of the pointers in the RAM buffer controller which are not currently selected, and for allowing software requests for read or write access by the node processor. Further, apparatus and a method are utilized for recording status and length information at the end of a packet instead of in front thereof and for allowing any incoming packet to be flushed without saving status information or to be flushed while saving its status information.
摘要:
There is disclosed herein an interface for a peripheral to allow the peripheral to communicate with either synchronous or asynchronous systems. The interface includes a series of flip-flops coupled in a chain to take the combined data strobe and chip select signals from an asynchronous system such as a microprocessor and synchronize the combined signal with the peripheral clock and convert the signal to a pulse. This pulse then emulates the instruction enable pulse received from synchronous microprogrammed systems. If the peripheral is operating in a synchronous sytem, the instruction enable signal from the system is gated through a multiplexer controlled by a mode signal which indicates which mode is active so as to act as the peripheral instruction enable signal. An AND gate gates one bit of the instruction bus with the ANDed chip select and data strobe signals to create a read/write control signal for internal use in controlling the data bus drivers.
摘要:
There is disclosed herein a CRC calculation circuit which can calculate CRC checkbits on 8 bits of raw input data per cycle of a byte clock. The calculation apparatus used 8 rows of shifting links with the inputs of each row coupled to the data outputs of the preceding row. Each shifting link shifts its input bit one bit position toward the most significant bit, and selected shifting links perform an exclusive-OR operation between their input bits and the output of an input exclusive-OR gate which exclusive-OR's one input bit with one of the bits in the most significant byte of the checksum register. A byte wide output bus is used to access the final checkbits from the checksum register by disabling the array of shifting links during the output cycles so that the bytes of CRC data can be shifted into position through the array one byte per each cycle of the byte clock. Preset logic for forcing all logic 1's into the data inputs of the first row of shifting links is provided such the machine can be preset during the first clock cycle of the CRC calculation. Several different architectures are disclosed for allowing separate calculation of CRC bits on a header packet and a data packet where the CRC bits on the data packet may be calculated on the data alone or the data plus the header and the CRC bits for the header. Logic for allowing CRC calculation to be performed on all bytes of a message while excluding some selected number of bits in the first byte is also disclosed.
摘要:
Methods and apparatus for embedding a control channel in a high speed serial interconnect having multiple data lanes. Operational aspects of the interconnect are controlled via use of control channel data that is sent over one or more of the data lanes on a periodic basis. A link state cycle is employed that includes a link control period during which control information is transferred over the interconnect and a link control interval between link control periods during which other links states are implemented, such as for transferring data or operating the link in a low power state. The link state cycles at transmitter and receiver ports are synchronized to account for link transmit latencies, and the timing of link state cycles corresponding to a bidirectional exchange of link control information may be configured to support an overlapping implementation or to facilitate a request/response link control protocol.