Abstract: A shared address space on a compute node stores data received from a network and data to transmit to the network. The shared address space includes an application buffer that can be directly operated upon by a plurality of processes, for instance, running on different cores on the compute node. A shared counter is used for one or more of signaling arrival of the data across the plurality of processes running on the compute node, signaling completion of an operation performed by one or more of the plurality of processes, obtaining reservation slots by one or more of the plurality of processes, or combinations thereof.

Abstract: Implementation primitives for concurrent array-based stacks, queues, double-ended queues (deques) and wrapped deques are provided. In one aspect, each element of the stack, queue, deque or wrapped deque data structure has its own ticket lock, allowing multiple threads to concurrently use multiple elements of the data structure and thus achieving high performance. In another aspect, new synchronization primitives FetchAndIncrementBounded (Counter, Bound) and FetchAndDecrementBounded (Counter, Bound) are implemented. These primitives can be implemented in hardware and thus promise a very fast throughput for queues, stacks and double-ended queues.

Abstract: A parallel computer system includes a plurality of compute nodes. Each of the compute nodes includes at least one processor, at least one memory, and a direct memory address engine coupled to the at least one processor and the at least one memory. The system also includes a network interconnecting the plurality of compute nodes. The network operates a global message-passing application for performing communications across the network. Local instances of the global message-passing application operate at each of the compute nodes to carry out local processing operations independent of processing operations carried out at another one of the compute nodes. The direct memory address engines are configured to interact with the local instances of the global message-passing application via injection FIFO metadata describing an injection FIFO in a corresponding one of the memories.

Abstract: A method and apparatus for managing coherence between two processors of a two processor node of a multi-processor computer system. Generally the present invention relates to a software algorithm that simplifies and significantly speeds the management of cache coherence in a message passing parallel computer, and to hardware apparatus that assists this cache coherence algorithm. The software algorithm uses the opening and closing of put/get windows to coordinate the activated required to achieve cache coherence. The hardware apparatus may be an extension to the hardware address decode, that creates, in the physical memory address space of the node, an area of virtual memory that (a) does not actually exist, and (b) is therefore able to respond instantly to read and write requests from the processing elements.

Abstract: Methods, apparatus, and products are disclosed for replenishing data descriptors in a Direct Memory Access (‘DMA’) injection first-in-first-out (‘FIFO’) buffer that include: determining, by a messaging module on an origin compute node, whether a number of data descriptors in a DMA injection FIFO buffer exceeds a predetermined threshold, each data descriptor specifying an application message for transmission to a target compute node; queuing, by the messaging module, a plurality of new data descriptors in a pending descriptor queue if the number of the data descriptors in the DMA injection FIFO buffer exceeds the predetermined threshold; establishing, by the messaging module, interrupt criteria that specify when to replenish the injection FIFO buffer with the plurality of new data descriptors in the pending descriptor queue; and injecting, by the messaging module, the plurality of new data descriptors into the injection FIFO buffer in dependence upon the interrupt criteria.

Abstract: A low latency memory system access is provided in association with a weakly-ordered multiprocessor system. Bach processor in the multiprocessor shares resources, and each shared resource has an associated lock within a locking device that provides support for synchronization between the multiple processors in the multiprocessor and the orderly sharing of the resources. A processor only has permission to access a resource when it owns the lock associated with that resource, and an attempt by a processor to own a lock requires only a single load operation, rather than a traditional atomic load followed by store, such that the processor only performs a read operation and the hardware locking device performs a subsequent write operation rather than the processor. A simple prefetching for non-contiguous data structures is also disclosed.

Abstract: A parallel computer system is constructed as a network of interconnected compute nodes to operate a global message-passing application for performing communications across the network. Each of the compute nodes includes one or more individual processors with memories which run local instances of the global message-passing application operating at each compute node to carry out local processing operations independent of processing operations carried out at other compute nodes. Each compute node also includes a DMA engine constructed to interact with the application via Injection FIFO Metadata describing multiple Injection FIFOs where each Injection FIFO may containing an arbitrary number of message descriptors in order to process messages with a fixed processing overhead irrespective of the number of message descriptors included in the Injection FIFO.

Abstract: A method for passing remote messages in a parallel computer system formed as a network of interconnected compute nodes includes that a first compute node (A) sends a single remote message to a remote second compute node (B) in order to control the remote second compute node (B) to send at least one remote message. The method includes various steps including controlling a DMA engine at first compute node (A) to prepare the single remote message to include a first message descriptor and at least one remote message descriptor for controlling the remote second compute node (B) to send at least one remote message, including putting the first message descriptor into an injection FIFO at the first compute node (A) and sending the single remote message and the at least one remote message descriptor to the second compute node (B).

Abstract: Methods, compute nodes, and computer program products are provided for direct memory access (‘DMA’) transfer completion notification. Embodiments include determining, by an origin DMA engine on an origin compute node, whether a data descriptor for an application message to be sent to a target compute node is currently in an injection first-in-first-out (‘FIFO’) buffer in dependence upon a sequence number previously associated with the data descriptor, the total number of descriptors currently in the injection FIFO buffer, and the current sequence number for the newest data descriptor stored in the injection FIFO buffer; and notifying a processor core on the origin DMA engine that the message has been sent if the data descriptor for the message is not currently in the injection FIFO buffer.

Abstract: A massively parallel supercomputer of petaOPS-scale includes node architectures based upon System-On-a-Chip technology, where each processing node comprises a single Application Specific Integrated Circuit (ASIC) having up to four processing elements. The ASIC nodes are interconnected by multiple independent networks that optimally maximize the throughput of packet communications between nodes with minimal latency. The multiple networks may include three high-speed networks for parallel algorithm message passing including a Torus, collective network, and a Global Asynchronous network that provides global barrier and notification functions. These multiple independent networks may be collaboratively or independently utilized according to the needs or phases of an algorithm for optimizing algorithm processing performance. The use of a DMA engine is provided to facilitate message passing among the nodes without the expenditure of processing resources at the node.

Abstract: Executing application function calls in response to an interrupt including creating a thread; receiving an interrupt having an interrupt type; determining whether a value of a semaphore represents that interrupts are disabled; if the value of the semaphore represents that interrupts are not disabled: calling, by the thread, one or more preconfigured functions in dependence upon the interrupt type of the interrupt; yielding the thread; and if the value of the semaphore represents that interrupts are disabled: setting the value of the semaphore to represent to a kernel that interrupts are hard-disabled; and hard-disabling interrupts at the kernel.

Abstract: An apparatus and method for capturing data packets for analysis on a network computing system includes a sending node and a receiving node connected by a bi-directional communication link. The sending node sends a data transmission to the receiving node on the bi-directional communication link, and the receiving node receives the data transmission and verifies the data transmission to determine valid data and invalid data and verify retransmissions of invalid data as corresponding valid data. A memory device communicates with the receiving node for storing the invalid data and the corresponding valid data. A computing node communicates with the memory device and receives and performs an analysis of the invalid data and the corresponding valid data received from the memory device.

Abstract: A parallel computer system is constructed as a network of interconnected compute nodes. Each of the compute nodes includes at least one processor, a memory and a DMA engine. The DMA engine includes a processor interface for interfacing with the at least one processor, DMA logic, a memory interface for interfacing with the memory, a DMA network interface for interfacing with the network, injection and reception byte counters, injection and reception FIFO metadata, and status registers and control registers. The injection FIFOs maintain memory locations of the injection FIFO metadata memory locations including its current head and tail, and the reception FIFOs maintain the reception FIFO metadata memory locations including its current head and tail. The injection byte counters and reception byte counters may be shared between messages.

Abstract: Embodiments of the invention may be used to manage message queues in a parallel computing environment to prevent message queue deadlock. A direct memory access controller of a compute node may determine when a messaging queue is full. In response, the DMA may generate an interrupt. An interrupt handler may stop the DMA and swap all descriptors from the full messaging queue into a larger queue (or enlarge the original queue). The interrupt handler then restarts the DMA. Alternatively, the interrupt handler stops the DMA, allocates a memory block to hold queue data, and then moves descriptors from the full messaging queue into the allocated memory block. The interrupt handler then restarts the DMA. During a normal messaging advance cycle, a messaging manager attempts to inject the descriptors in the memory block into other messaging queues until the descriptors have all been processed.

Abstract: A system and method for enabling high-speed, low-latency global tree network communications among processing nodes interconnected according to a tree network structure. The global tree network enables collective reduction operations to be performed during parallel algorithm operations executing in a computer structure having a plurality of the interconnected processing nodes. Router devices are included that interconnect the nodes of the tree via links to facilitate performance of low-latency global processing operations at nodes of the virtual tree and sub-tree structures. The global operations performed include one or more of: broadcast operations downstream from a root node to leaf nodes of a virtual tree, reduction operations upstream from leaf nodes to the root node in the virtual tree, and point-to-point message passing from any node to the root node.

Abstract: A method and apparatus for managing coherence between two processors of a two processor node of a multi-processor computer system. Generally the present invention relates to a software algorithm that simplifies and significantly speeds the management of cache coherence in a message passing parallel computer, and to hardware apparatus that assists this cache coherence algorithm. The software algorithm uses the opening and closing of put/get windows to coordinate the activated required to achieve cache coherence. The hardware apparatus may be an extension to the hardware address decode, that creates, in the physical memory address space of the node, an area of virtual memory that (a) does not actually exist, and (b) is therefore able to respond instantly to read and write requests from the processing elements.

Abstract: A method and apparatus for re-utilizing partially failed compute resources in a massively parallel super computer system. In the preferred embodiments the compute node comprises a number of clock domains that can be enabled separately. When an error in a compute node is detected, and the failure is not in network communication blocks, a clock enable circuit enables the clocks to the network communication blocks only to allow the partially failed compute node to be re-utilized as a network resource. The computer system can then continue to operate with only slightly diminished performance and thereby improve performance and perceived overall reliability.

Abstract: A novel massively parallel supercomputer of hundreds of teraOPS-scale includes node architectures based upon System-On-a-Chip technology, i.e., each processing node comprises a single Application Specific Integrated Circuit (ASIC). Within each ASIC node is a plurality of processing elements each of which consists of a central processing unit (CPU) and plurality of floating point processors to enable optimal balance of computational performance, packaging density, low cost, and power and cooling requirements. The plurality of processors within a single node may be used individually or simultaneously to work on any combination of computation or communication as required by the particular algorithm being solved or executed at any point in time. The system-on-a-chip ASIC nodes are interconnected by multiple independent networks that optimally maximizes packet communications throughput and minimizes latency.

Abstract: A method and apparatus for supporting cache coherency in a multiprocessor computing environment having multiple processing units, each processing unit having one or more local cache memories associated and operatively connected therewith. The method comprises providing a snoop filter device associated with each processing unit, each snoop filter device having a plurality of dedicated input ports for receiving snoop requests from dedicated memory writing sources in the multiprocessor computing environment. Each of the memory writing sources is directly connected to the dedicated input ports of all other snoop filter devices associated with all other processing units in a point-to-point interconnect fashion.

Abstract: Class network routing is implemented in a network such as a computer network comprising a plurality of parallel compute processors at nodes thereof. Class network routing allows a compute processor to broadcast a message to a range (one or more) of other compute processors in the computer network, such as processors in a column or a row. Normally this type of operation requires a separate message to be sent to each processor. With class network routing pursuant to the invention, a single message is sufficient, which generally reduces the total number of messages in the network as well as the latency to do a broadcast. Class network routing is also applied to dense matrix inversion algorithms on distributed memory parallel supercomputers with hardware class function (multicast) capability. This is achieved by exploiting the fact that the communication patterns of dense matrix inversion can be served by hardware class functions, which results in faster execution times.