Abstract: One aspect described in this application provides an apparatus that includes an external tray hose with an integrated booster pump for cooling a computing device. The booster pump can boost fluid flow and fluid pressure of coolant fluid pumped toward the computing device on a server tray. The booster pump includes a stator mounted around a booster pump housing with a bore. The stator includes one or more C-shaped lamination stacks for supporting coil windings. Energized coil windings generate a varying magnetic field for rotating an impeller within a housing bore to pump fluid along the housing bore of the booster pump. The apparatus includes a monitor module to monitor a set of sensors to obtain information about one or more of fluid temperature, fluid pressure, and device temperature; and a control module to adjust performance of the booster pump based on the information obtained from the set of sensors.

Abstract: A memory cooler includes a unitary thermal transfer device and a pair of endcaps. The unitary thermal transfer device includes heat transfer tubes, a first end block, a second end block, an inlet chamber, an outlet chamber, an inlet, and an outlet. The first and second end blocks are structurally integrated with each of the heat transfer tubes. The inlet and outlet chambers are partially defined by either the first end block or the second end block. Each of the inlet and outlet chambers are fluidly coupled with the respective liquid flow channel of the at least one heat transfer tube. The respective flow channels to which the inlet and outlet chambers are coupled may be the same or different to define either a direct or a serpentine flow path. Each endcap is affixed to a respective one of the first end block and the second end block to define, in conjunction with the first end block and second end block, the inlet chamber and the outlet chamber.

Abstract: A process includes accessing a query that is associated with joining a fact table with a plurality of dimension tables. The fact table includes a primary key that includes a set of columns, which are covered by the dimension tables. The process includes determining a query plan for processing the query. The query plan has an associated join order. Determining the query plan includes determining a plurality of sets of the plurality of dimension tables, where each set includes dimension tables, which cover the set of columns of the primary key. Determining the query plan includes evaluating costs that are associated with joining subsets of the plurality of sets with the fact table. Based on the costs, a given subset of the plurality of subsets is selected. Determining the query plan includes constraining the join order based on the given subset so that the dimension tables of the given subset are joined before the fact table.

Abstract: In some examples, a system partitions a shared memory address space of a shared memory among a plurality of processing entities into a plurality of memory partitions, where a respective memory partition is associated with a respective processing entity. A first processing entity forwards, to a second processing entity, a first data operation, based on a determination by the first processing entity that the first data operation is to be applied to data for a memory partition associated with the second processing entity. The second processing entity applies the first data operation that includes writing data of the first data operation to the memory partition associated with the second processing entity using a non-atomic operation.

Abstract: The system determines a first and a second snapshot of memory usage by processes in a computing system based on a predetermined time interval, wherein a respective snapshot indicates, for a respective process, memory usage corresponding to memory types including private memory and shared memory. The system computes, for processes common to the first and the second snapshots, a likelihood of memory leak for a process and memory type based on: an increase in an amount of heap memory and private memory; an increase in an amount of shared memory; and an increase in a total amount of private and shared memory. The system compares the computed likelihood for the process with a predetermined threshold. Responsive to determining that the computed likelihood for the process exceeds the predetermined threshold, the system performs a corrective action to address memory leak associated with the process.

Abstract: Systems and methods are provided for implementing swarm learning while using blockchain technology and election/voting mechanisms to ensure data privacy. Nodes may train local instances of a machine learning model using local data, from which parameters are derived or extracted. Those parameters may be encrypted and persisted until a merge leader is elected that can merge the parameters using a public key generated by an external key manager. A decryptor that is not the merge leader can be elected to decrypt the merged parameter using a corresponding private key, and the decrypted merged parameter can then be shared amongst the nodes, and applied to their local models. This process can be repeated until a desired level of learning has been achieved. The public and private keys are never revealed to the same node, and may be permanently discarded after use to further ensure privacy.

Abstract: Implementations of the present disclosure relate to a method for decoding error correction. The method comprises detecting a failure of decoding a received frame. After the failure of decoding the received frame is detected, a type of the received frame is determined based on a probability that the received frame follows a prior frame of the received frame in a frame sequence. The method further comprises obtaining a template corresponding to the type of the received frame, and decoding the received frame based on the fixed values in the template. The template includes fixed values corresponding to the type of the received frame. With these implementations, correction ability of the decoding can be obviously improved with assistance of the constructed template.

Abstract: A network interface controller (NIC) capable of facilitating efficient memory address translation is provided. The NIC can be equipped with a host interface, a cache, and an address translation unit (ATU). During operation, the ATU can determine an operating mode. The operating mode can indicate whether the ATU is to perform a memory address translation at the NIC. The ATU can then determine whether a memory address indicated in the memory access request is available in the cache. If the memory address is not available in the cache, the ATU can perform an operation on the memory address based on the operating mode.

Abstract: A process includes, in a computer system, acquiring a first measurement that corresponds to a software container. Acquiring the measurement includes a hardware processor of the computer system measuring a given layer of a plurality of layers of layered file system structure corresponding to the software container. The given layer includes a plurality of files, and the first measurement includes a measurement of the plurality of files. The process includes storing the first measurement in a secure memory of the computer system. A content of the secure memory is used to verify an integrity of the software container.

Abstract: A network interface controller (NIC) capable of efficient operation management for host accelerators is provided. The NIC can be equipped with a host interface and triggering logic block. During operation, the host interface can couple the NIC to a host device. The triggering logic block can obtain, via the host interface from the host device, an operation associated with an accelerator of the host device. The triggering logic block can determine whether a triggering condition has been satisfied for the operation based on an indicator received from the accelerator. If the triggering condition has been satisfied, the triggering logic block can obtain a piece of data generated from the accelerator from a memory location and execute the operation using the piece of data.

Abstract: Techniques and systems for a semiconductor laser, namely a grating-coupled surface-emitting (GCSE) comb laser, having thermal management for facilitating dissipation of heat, integrated thereon. The thermal management is structured in manner that prevents deformation or damage to the GCSE laser chips included in the semiconductor laser implementation. The disclosed thermal management elements integrated in the laser can include: heat sinks; support bars; solder joints; thermal interface material (TIM); silicon vias (TSV); and terminal conductive sheets. Support bars, for example, having the GCSE laser chip positioned between the bars and having a height that is higher than a thickness of the GCSE laser chip. Accordingly, the heat sink can be placed on top of the support bars such that heat is dissipated from the GCSE laser chip, and the heat sink is separated from directed contact with the GCSE laser chip due to the height of the support bars.

Abstract: Examples described herein relate to a manageability controller for controlling a display of a screen video. The manageability controller may receive screen video data from a hypervisor running on a host operating system (OS) that is executable by a main processing resource separate from the manageability processing resource. The screen video data may include a host OS screen video data corresponding to the host OS, a virtual machine (VM) screen video data corresponding to a VM running on the hypervisor, or both. Further, the manageability controller may store the host OS screen video data or the VM screen video data in a physical video memory based on a screen selection input.

Abstract: A system comprising processing circuitry a memory storing instructions that cause the system to detect a code change to source code included in a code repository, identify a relationship between the code change and an associated product feature, determine one or more dependent product features impacted by the code change, select a set of test cases including a subset of test cases related to the associated product feature and a subset of test cases related to the one or more dependent product features, execute the set of test cases, and update the code-to-feature mapping using results of executing the set of test case.

Abstract: Example techniques for identification of diagnostic messages corresponding to exceptions are described. A determination model may determine whether a set of diagnostic messages generated based on analysis of a source code includes a diagnostic message that likely corresponds to an exception. The determination may be used to identify a set of diagnostic messages including the diagnostic message that likely corresponds to an exception.

Abstract: Examples include a blockchain system associated with a distributed file system, Some examples include retrieval, from the blockchain system, a latest block corresponding to a block signature of a reference block that corresponds to a file stored in the distributed file system. The latest block includes information identifying at least one of a latest location of the file, a signature of metadata of the file, or a tag associated with the file.

Abstract: In exemplary aspects, a golden data structure can be used to validate the stability of machine learning (ML) models and weights. The golden data structure includes golden input data and corresponding golden output data. The golden output data represents the known correct results that should be output by a ML model when it is run with the golden input data as inputs. The golden data structure can be stored in a secure memory and retrieved for validation separately or together with the deployment of the ML model for a requested ML operation. If the golden data structure is used to validate the model and/or weights concurrently with the performance of the requested operation, the golden input data is combined with the input data for the requested operation and run through the model. Relevant outputs are compared with the golden output data to validate the stability of the model and weights.

Abstract: A method for securing a plurality of compute nodes includes authenticating a hardware architecture of each of a plurality of components of the compute nodes. The method also includes authenticating a firmware of each of the plurality of components. Further, the method includes generating an authentication database comprising a plurality of authentication descriptions that are based on the authenticated hardware architecture and the authenticated firmware. Additionally, a policy for securing a specified subset of the plurality of compute nodes is implemented by using the authentication database.

Abstract: In some examples, a system identifies sub-portions of a database query, assigns identifiers to the identified sub-portions, and adds the identifiers to a data structure. The system generates a fingerprint representing the database query based on applying a fingerprint function on the data structure including the identifiers.

Abstract: Disclosed is a system and method for the monitoring and authorization of an optimization device in a network. In exemplary embodiments, an optimization device transmits an authorization request message to a portal to receive authorization to operate. The portal transmits an authorization response message to the optimization device with capability parameters for operation of the device, including at least one expiration parameter for the authorization. The optimization device sends updated authorization request messages to the portal with its device usage information, such that the portal can dynamically monitor the optimization device and continue to authorize its operation.

Abstract: Examples include aggregating logs, where each of the logs is associated with a workflow instance. Each log includes information indicative of an event occurring during the workflow instance. Further, examples include assigning, based on user intent of the workflow instance, a workflow name to each log, where the user intent is indicative of an outcome of execution of the workflow instance and assigning an instance identifier to each log, where the instance identifier corresponds to the workflow instance. Further, identifying a subset of the plurality of logs having an identical workflow name and an identical instance identifier, associating a tracking identifier to the subset, and creating an index of processed logs, wherein each processed log in the index includes the tracking identifier. Further, analyzing the index of processed logs based on a set of rules and identifying, based on the analysis, an error in execution of each the workflow instance.