Hewlett Packard Enterprise Development LP

Ryan Dean Menhusen

Darel N Emmot

Daniel William Dauwe

Harumi Kuno

Duncan Roweth

Alan Lynn Davis

Sai Rahul Chalamalasetti

Sergey Serebryakov

Torsten Wilde

John L Byrne

Todd Austin Carrington

Vipin Kumar Kukkala

A system for facilitating sender-side congestion control is provided. During operation, the system, on a sender node, can determine the utilization of a buffer at a last-hop switch to a receiver node based on in-flight packets to the receiver node. The receiver node can be reachable from the sender node via the last-hop switch. The system can then determining a fraction of available space in the buffer for packets from the sender node based on the utilization of the buffer. Subsequently, the system can determine whether the fraction of the available space in the buffer can accommodate a next packet from the sender node while avoiding congestion at the buffer at the receiver node. If the fraction of the available space in the buffer can accommodate the next packet, the system can allow the sender node to send the next packet to the receiver node.

Method and system for dynamic quota-based congestion management

A system for facilitating sender-side granular congestion control is provided. During operation, the first and second processes of an application can run on sender and receiver nodes, respectively. A first buffer on the sender node can be allocated to the first process. For the first process, the system can then identify a second buffer at a last-hop switch of the receiver node. The system can determine, based on in-flight packets, the utilization of the second buffer. The system can also determine a fraction of available space in the second buffer for packets from the first buffer based on the utilization. Subsequently, the system can determine whether the fraction of the available space can accommodate the next packet from the first buffer. If the fraction of the available space can accommodate the next packet, the system can allow the first process to send the next packet to the second process.

Method and system for granular dynamic quota-based congestion management

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.

Memory partitions for processing entities

In some examples, a system includes a processing entity and a memory to store data arranged in a plurality of bins associated with respective key values of a key. The system includes a cache to store cached data elements for respective accumulators that are updatable to represent occurrences of the respective key values of the key, where each accumulator corresponds to a different bin of the plurality of bins, and each cached data element has a range that is less than a range of a corresponding bin of the plurality of bins. Responsive to a value of a given cached data element as updated by a given accumulator satisfying a criterion, the processing entity is to cause an aggregation of the value of the given cached data element with a bin value in a respective bin.

Accumulators corresponding to bins in memory

One embodiment provides a system and method for predicting network power usage associated with workloads. During operation, the system configures a simulator to simulate operations of a plurality of network components, which comprises embedding one or more event counters in each simulated network component. A respective event counter is configured to count a number of network-power-related events. The system collects, based on values of the event counters, network-power-related performance data associated with one or more sample workloads applied to the simulator; and trains a machine-learning model with the collected network-power-related performance data and characteristics of the sample workloads as training data 1, thereby facilitating prediction of network-power-related performance associated with a to-be-evaluated workload.

System and method for predicting power usage of network components

Systems and methods are provided for accurately simulating memory operations of a multi-compute-engine system, such as a multi-core system. Simulation speed can be increased by consolidation location and state information associated with data stored in one or more caches of a simulated cache hierarchy. This consolidation of information can be reflected in a single cache line map or flat cache. Accordingly, searches for data (and copies of the data) in each and every cache of the simulated cache hierarchy can be performed fast and with greater efficiency than conventional simulation systems that operate using sequential, cache-by-cache searching, while maintaining data coherency.

Growing community of inventors

Ft. Collins, CO, United States of America

Ryan Dean Menhusen