International Business Machines Corporation

Joseph M Hoke

Thomas Anthony Gregg

Frank David Ferraiolo

Chin Lee

Samir Kirit Patel

Albert Ing

Kulwant M Pandey

Tin-Chee Lo

Robert R Livolsi

Daniel Francis Casper

Jay Robert Herring

Edward W Chencinski

Patrick John Sugrue

Robert James Lynch

Steven Louis Makow

Paul W Bond

Andrew Z Muszynski

David B Fox

John Michael Yarolin

Initialization of a bidirectional, self-timed parallel interface with capacitive coupling is provided. The self-timed interface includes master and slave nodes connected by a parallel bus comprising multiple AC differential wire pairs. The initialization includes automatically testing at least one wire pair of the multiple AC differential wire pairs for conductivity failure, wherein the testing is responsive to a link reset signal of a first frequency. The automatically testing includes employing a link test signal of a second frequency to test the at least one wire pair of the multiple AC differential wire pairs. The second frequency is a lower frequency than a third, operational signal frequency of the self-timed parallel interface, and the first frequency and the second frequency comprise different frequencies. An initialization testing and handshake approach between the master node and slave node is also provided.

Initialization of a bidirectional, self-timed parallel interface with automatic testing of AC differential wire pairs

A method is disclosed for instructing a computing system to allocate a trace array from an original cache memory, where the method includes dividing the original cache memory into a reduced-size cache memory and a trace array, permitting storage of trace signal data into the trace array, and permitting retrieval of the trace signal data from the trace array.

Method for using a portion of the system cache as a trace array

An exemplary embodiment of the present invention is a method for transmitting data among processors over a plurality of parallel data lines and a clock signal line. A receiver processor receives both data and a clock signal from a sender processor. At the receiver processor a bit of the data is phased aligned with the transmitted clock signal. The phase aligning includes selecting a data phase from a plurality of data phases in a delay chain and then adjusting the selected data phase to compensate for a round-off error. Additional embodiments include a system and storage medium for transmitting data among processors over a plurality of parallel data lines and a clock signal line.

Method and system for selecting data sampling phase for self timed interface logic

A transmission line for carrying digital signals is provided with auxiliary line drivers at the receiver end, which are used in combination with the data line driver at the transmitting end to apply a sequence of sets of test voltages onto the line. The voltage resulting at the receiver input is used to detect open signal conductors and/or shorts in the receiver.

Method and apparatus for testing a signal line

Frequency differences between differing clock sources are compensated for by an adaptive filtering mechanism. An amount of frequency drift between two clock sources is determined. Then, based on that amount of frequency drift, a filtering value is selected to be used in tracking the frequency drift. If the frequency drift is determined to be large, then a minimum filtering value is selected. However, if it is determined to be small, then a maximum filtering value is selected. The selected filtering value is used to adjust the address(es) of one or more data bits being transmitted and received using the two clock sources, such that the frequency drift is properly tracked.

Adaptive filtering method and apparatus to compensate for a frequency

A serial data stream is recovered using a local clock, which is asynchronous to the clock used to transmit the serial data. The incoming serial data stream is phase shifted or delayed by a digital phase-locked loop so that it may be reliably sampled by the local clock. The DPLL samples the serial data stream and captures data on both the rising and falling edges of the local clock employing three edge detectors. This partitions the data stream into two bit samples, which the DPLL presents to a deserializer. The deserializer converts the serial data to parallel data and assembles the received data back into data bytes. The deserializer also generates a received byte clock used for presenting the parallel data to, for example, the ESCON channel logic.

Method and apparatus for recovering a serial data stream using a local

A system and method for asynchronously receiving data blocks, in parallel, across multiple fibers in a serial manner. Frame groups are provided as a mechanism to transmit associated data serially on each fiber and tie the data being transmitted together. The frame groups do not have sequence numbers, therefore, the receiver determines which frames are part of a frame group by the arrival times of the individual frames. The transceivers for each member of the parallel bus examine the received bit stream to extract frames and continuous sequences. For each member of the parallel bus there are independent receive buffers, and these buffers are controlled by independent states. The states inhibit erroneously generated frames from corrupting the contents of the receive buffers and inhibit the loading of the buffers after errors on the link. These states also control the loading of the receive buffers after retransmission of a buffer area. The information from the individual frames of the frame group is assembled in the proper sequence by another element in the channel. This element also suppresses notification of the intermediate frame groups when multiple frame groups are used to transmit a buffer area.

System for identifying communication sequences transmitted across

A system and method for asynchronously transmitting data blocks, in parallel, across multiple fibers in a serial manner. Frame groups are provided as a mechanism to transmit associated data serially on each fiber and tie the data being transmitted together. The frame groups can be prematurely ended on any 256 byte block boundary for several purposes, and all the frames of a given group must contain the same number of information field data words. Allowing frame groups to be ended on arbitrary block boundaries allows their transmission to start before all of the information field for the frame group has been received from a shared main processor storage. This capability of ending frame groups also allows high priority frame groups to interrupt the transmission of a relatively long data frame. Finally, the capability to end the frame group protects the information field of a stalled frame group since the CRC is sent and the idle sequence is resumed. Using multiple, independently clocked transmitters improves the fault tolerance of the link but also makes it difficult to end all frame of the frame group on the same block boundary. This problem is also solved by keeping track of the data sent to the multiple transmit buffers.

Suspending, resuming, and interleaving frame-groups

A system and method for asynchronously transmitting data across fibers in a serial manner. Frames are provided as a mechanism to transmit associated data serially and tie the data being transmitted to a particular buffer set. Each buffer set maintains a state that keeps track of the progress and sequence of received frames. When transmission errors occur in the frames, the errors may affect header bits in the frame that identify the buffer set and the frame type. In this case, the entire frame is lost and the operation usually results in a timeout of the operation. By using this state information some of these lost frames can be recovered avoiding the lengthier timeout recovery procedure.

Recovery of lost frames in a communication link

A system and method for asynchronously transmitting data across fibers in a serial manner. Frames are provided as a mechanism to transmit associated data serially and tie the data being transmitted to a particular buffer set. Each outstanding request for each buffer set is individually timed to detect lost frames, and each buffer set maintains a state that keeps track of the progress and sequence of received frames. When transmission errors occur in the frames, the errors may affect only the information field and there is enough information in the header to identify the frame. A request can then be sent back to the other end of the fibers to retransmit the frame. In some instances, the frame cannot be retransmitted because the content of the data may have been changed by other operations in the systems. To speed recovery in these situations, the requester of the retransmitted frame sets a much shorter timer while waiting for the response since the retransmitted frame will be sent quickly, if at all.

Shortened timeout period during frame retry in a communication link

A system and method for asynchronously transmitting data blocks, in parallel, across multiple fibers in a serial manner. Frame groups are provided as a mechanism to transmit associated data serially on each fiber and tie the data being transmitted together. The frame groups do not have sequence numbers, therefore, the receiver determines which frames are part of a frame group by the arrival times of the individual frames. The transceivers for each member of the parallel bus asynchronously achieve synchronism from either end of the fiber. A receiver of a transceiver determines a relative skew on a conductor of a link by performing skew measurements on a series of calibration frames sent by a transmitter on another end of said link. A skew monitor performs skew measurements and includes a plurality of skew measurement registers for measuring the relative skew between frames of a frame group, each register having an input from a skew counter, and outputs coupled to a microprocessor controlling the monitor. Each register gates in the value in the skew counter at the time a link control word is received. The receiver can determine that one or more frames of the calibration frames have been lost and have not exceeded the maximum measureable skew limits.

Skew measurement for receiving frame-groups

A system and method for asynchronously transmitting data across fibers in a serial manner. Frames are provided as a mechanism to transmit associated data serially and tie the data being transmitted to a particular buffer set. Each buffer set maintains a state that keeps track of the progress and sequence of received frames. When transmission errors occur in the frames, the errors may affect only the information field and there is enough information in the header to identify the frame. A request can then be sent back to the other end of the fibers to retransmit the frame. Certain information in the frame header is not used to identify the frame, but is used to control the transfer of frames and identify lost frames. When a frame information field is damaged by transmission errors, this header information is retained by the receiver of the damaged frame. Retaining this information at the receiver of the damaged frame frees the sender for storing it during the transmission of each frame. Freeing the sender improves performance by removing steps for the normal error free transmission of frames.

Growing community of inventors

Millerton, NY, United States of America

Joseph M Hoke