Abstract: In one example, a memory network may control access to a shared memory that is by multiple compute nodes. The memory network may control the access to the shared memory by receiving a memory access request originating from an application executing on the multiple compute nodes and determining a priority for processing the memory access request. The priority determined by the memory network may correspond to a memory address range in the memory that is specifically used by the application.

Abstract: A system management command is stored in a management partition of a global memory by a first node of a multi-node computing system. The global memory is shared by each node of the multi-node computing system. In response to an indication to access the management partition, the system management command is accessed from the management partition by a second node of the multi-node computing system. The system management command is executed by the second node. Executing the system management command includes managing the second node.

Abstract: According to an example, an instruction to run a kernel of an application on an apparatus having a first processing unit integrated with a second processing unit may be received. In addition, an application profile for the application at a runtime of the application kernel on the second processing unit may be created, in which the application profile identifies an affinity of the application kernel to be run on either the first processing unit or the second processing unit, and identifies a characterization of an input data set of the application. The application profile may also be stored in a data store.

Abstract: A system comprises a plurality of functional units powered via a power source. The system further comprises a first functional unit and a second functional unit, wherein the second functional unit is to promote the first functional unit to a management unit based on a management requirement of the system. The management unit is to administrate operations of the system. Once the first functional unit is promoted, the management unit is isolated from the functional units that were not promoted via a virtual network path and a power management unit.

Abstract: A system comprises a plurality of functional units powered via a power source. The system further comprises a first functional unit and a second functional unit, wherein the second functional unit is to promote the first functional unit to a management unit based on a management requirement of the system. The management unit is to administrate operations of the system. Once the first functional unit is promoted, the management unit is isolated from the functional units that were not promoted via a virtual network path and a power management unit.

Abstract: Performance of a computing system is improved by identifying and mitigating a bottleneck along a path that spans a storage system and a virtual machine causing the bottleneck. A mitigation action is selected and performed according to the bottleneck location. To identify a virtual machine involved in the bottleneck, end-to-end latency values connected with individual virtual machines are used, some of which are estimated using the presently disclosed techniques. Specifically, a backend storage latency from a specific virtual machine, and a flash virtualization platform, network, and queuing latency for the virtual machine are not conventionally observable, but are instead estimated using other readily available usage statistics.

Abstract: Performance of a computing system is improved by identifying storage clients that generate relatively large workloads and mitigating overall impact of the identified storage clients. Latency is monitored for different write and read I/O request block sizes. When latency for any request bock size increases above a predefined threshold for the request block size, diagnostics are performed to identify a storage client that generated an excessive workload or workloads comprising large blocks that caused the increased latency. A mitigation action can then be performed.

Abstract: The communication between integrated graphics processing units (GPUs) is disclosed. A first integrated GPU of a first computing device obtains a tuple pertaining to data to be transmitted to a second integrated GPU of a second computing device. The tuple comprises at least a length of the data. The first integrated GPU allocates a virtual address space to the data based on the length of the data, where the virtual address space has a plurality of virtual addresses. Further, a mapping table of a mapping between the plurality of virtual addresses and a plurality of bus addresses is provided by the first integrated GPU to a communication module of the first computing device to transmit the data, where the plurality of bus addresses indicate physical locations of the data.

Abstract: Performance of a computing system is improved by avoiding and/or eliminating overload conditions in storage systems. Performance utilization calculations are used to predict that a candidate placement configuration of storage resources within the storage systems will avoid overload conditions. The performance utilization calculations use performance profiles in conjunction with workload profiles to account for both storage system performance and actual workload traffic. Performance utilization calculations can also be used to report storage controller utilization information.

Abstract: In one example, a memory network may control access to a shared memory that is by multiple compute nodes. The memory network may control the access to the shared memory by receiving a memory access request originating from an application executing on the multiple compute nodes and determining a priority for processing the memory access request. The priority determined by the memory network may correspond to a memory address range in the memory that is specifically used by the application.

Abstract: According to an example, an instruction to run a kernel of an application on an apparatus having a first processing unit integrated with a second processing unit may be received. In addition, an application profile for the application at a runtime of the application kernel on the second processing unit may be created, in which the application profile identifies an affinity of the application kernel to be run on either the first processing unit or the second processing unit, and identifies a characterization of an input data set of the application. The application profile may also be stored in a data store.

Abstract: A computing node includes an active Non-Volatile Random Access Memory (NVRAM) component which includes memory and a sub-processor component. The memory is to store data chunks received from a processor core, the data chunks comprising metadata indicating a type of post-processing to be performed on data within the data chunks. The sub-processor component is to perform post-processing of said data chunks based on said metadata.

Abstract: Managing a network system includes determining metrics for a plurality of nodes in the network system, determining a plurality of zones including the plurality of nodes based on the metrics for the network system, and, for each zone of the plurality of zones, determining a computational architecture to be implemented for the zone based on the metrics for each node of the plurality of nodes in the zone.

Abstract: A system comprises a plurality of functional units powered via a power source. The system further comprises a first functional unit and a second functional unit, wherein the second functional unit is to promote the first functional unit to a management unit based on a management requirement of the system. The management unit is to administrate operations of the system. Once the first functional unit is promoted, the management unit is isolated from the functional units that were not promoted via a virtual network path and a power management unit.

Abstract: A system management command is stored in a management partition of a global memory by a first node of a multi-node computing system. The global memory is shared by each node of the multi-node computing system. In response to an indication to access the management partition, the system management command is accessed from the management partition by a second node of the multi-node computing system. The system management command is executed by the second node. Executing the system management command includes managing the second node.

Abstract: The communication between integrated graphics processing units (CPUs) is disclosed. A first integrated GPU of a first computing device obtains a tuple pertaining to data to be transmitted to a second integrated CPU of a second computing device. The tuple comprises at least a length of the data. The first integrated GPU allocates a virtual address space to the data based on the length of the data, where the virtual address space has a plurality of virtual addresses. Further, a mapping table of a mapping between the plurality of virtual addresses and a plurality of bus addresses is provided by the first integrated GPU to a communication module of the first computing device to transmit the data, where the plurality of bus addresses indicate physical locations of the data.

Abstract: Examples for performing processing tasks using an auxiliary processing unit are described. In an example, a computing system may include a processor to perform a plurality of processing tasks for each of a plurality of applications hosted by the computing system. An auxiliary processing task may be determined for an active application from the plurality of applications. The auxiliary processing tasks may be from among the plurality of processing tasks performed for the active application. Further, the processing code corresponding to the auxiliary processing task may be provided to the auxiliary processing unit of the computing system. The auxiliary processing unit may execute the processing code to perform corresponding auxiliary processing tasks and share a processing load with the processor.

Abstract: A method for cross-layer power management in a multi-layer system includes determining whether there is a service level violation for an application running on a hardware platform. Power consumption of the hardware platform is controlled in response to the service level violation.

Abstract: Log analysis can include transferring compiled log analysis code, executing log analysis code, and performing a log analysis on the executed log analysis code.

Abstract: A technique includes receiving a user input in an array-oriented database. The user input indicates a database operation and processing a plurality of chunks of data stored by the database to perform the operation. The processing in dudes selectively distributing the processing of the plurality of chunks between a first group of at least one central processing unit and a second group of at least one co-processor.