Abstract: A scheme is presented for storing data on disks in such a way that a checkpoint can easily be taken across several disks connected to different processors in a distributed or parallel computer. A checkpoint can be used to restore the entire disk system to a known state after one or more of the disks or processors fails. When a failure occurs, the disk system is restored to its state at the current checkpoint. The scheme allows significant saving in disk space by requiring that only the data modified since the last checkpoint be copied. The checkpointing algorithm is presented as part of the invention. The invention allows checkpointing of disk space independently of the use of the disk space, for example, in a file system.

Abstract: Method and system for managing process scheduling among multiple control contexts within a data processing environment. A given control context, comprising one of multiple control contexts, is assumed to be executing at a processing node within the environment. The approach includes processing a scheduling event; evaluating process scheduling from within the executing control context to determine a restart control context, which also comprises one of the multiple control contexts; and restarting executing data within the restart control context in response to the process scheduling occurring within the given control context. Context switching is employed only if process scheduling determines that the restart control context is other than the given control context. The technique is particularly useful in nonpreemptive parallel data processing environments.

Abstract: A message packet transmitter for transmitting a packet of electronic data signals onto a communication network without interruption. A first-in, first-out electronic memory has a transmit state in which an electronic data signal stored therein is output. The electronic memory has an idle state in which electronic data signals stored therein are not output. A message packet-in-transit identification circuit generates a packet-in-transit signal after the electronic memory outputs the first electronic data signal in a message packet. A no-packet-in-transit signal is generated after the electronic memory outputs the last electronic data signal in the message packet. A state controller maintains the electronic memory in the transmit state when the electronic memory stores at least a portion of a message packet, when the state controller receives a packet-in-transit signal, and when the state controller receives an interrupt-pending signal.

Abstract: An output window is created on the user's terminal screen when a parallel program is executing. This window displays an array of graphical elements (each preferably a small square area), which are partitioned into groups of one or more graphical elements per group, each partition representing a task or thread of the parallel program. These graphical elements are capable of assuming any one of several (or many) different graphical states (each of these states preferably being a different color for the graphical clement). A task running on a parallel processor system can set its associated graphical elements to different states (e.g., colors) during execution of the task generally through a special instruction in the task that specifics which graphical clement (of the graphical elements assigned to that task) should be set and to what state. When such an instruction is executed by a processor running that task, a message is sent to the module that controls the graphical element display at the user's terminal.