Google Inc.

Nutanix, Inc.

Other

Aip Semi, Inc.

Jianjun Wen

Xinyu Tang

Himanshu Shukla

Cong Liu

Henry Verheyen

Binny Sher Gill

Abhinay Nagpal

Weiheng Chen

Shuo Yang

A hardware accelerator that is efficient at performing computations related to a neural network. In one embodiment, the hardware accelerator includes a first data buffer that receives input data of a layer in the neural network and shift the input data slice by slice downstream. The hardware accelerator includes a second data buffer that receives kernel data of the layer in the neural network and shift the kernel data slice by slice downstream. The hardware accelerator includes a first input shift register that receives an input data slice from the first data buffer. The first input shift register may correspond to a two-dimensional shift register configured to shift values in the input data slice in x and y directions. The hardware accelerator includes a second input shift register that receives a kernel data slice from the second data buffer. A multiplication block performs convolution of the input and kernel data.

Hardware architecture for processing data in neural network

A method for time series analysis of time-oriented usage data pertaining to computing resources of a computing system. A method embodiment commences upon collecting time series datasets, individual ones of the time series datasets comprising time-oriented usage data of a respective individual computing resource. A plurality of prediction models are trained using portions of time-oriented data. The trained models are evaluated to determine quantitative measures pertaining to predictive accuracy. One of the trained models is selected and then applied over another time series dataset of the individual resource to generate a plurality of individual resource usage predictions. The individual resource usage predictions are used to calculate seasonally-adjusted resource usage demand amounts over a future time period. The resource usage demand amounts are compared to availability of the resource to form a runway that refers to a future time period when the resource is predicted to be demanded to its capacity.

Method for forecasting distributed resource utilization in a virtualization environment

Systems for computing cluster management. One embodiment commences upon receiving a set of observed workload parameters corresponding to one or more observable workloads that run in a computing cluster. While the workloads are running, workload stimulus and cluster response observations are taken and used to generate a workload resource usage predictive model based on mappings or correlations between the observable workloads parameters and observed resource usage measurements. A set of planned workloads are applied to the workload resource usage predictive model to predict a set of corresponding predicted resource usage demands. The predicted resource usage demands are then mapped to a set of recommended hardware to form resource deployment recommendations that satisfy at least some of the corresponding resource usage demands while also observing a set of hardware model compatibility constraints. The resource deployment recommendations that satisfy the set of hardware model compatibility constraints are displayed in a user interface.

Long-range distributed resource planning using workload modeling in hyperconverged computing clusters

A method performed by one or more processing devices includes obtaining a definition of a data structure; obtaining a definition of one or more operations to be performed on the data structure, the definition of the one or more operations being in accordance with a first syntax; receiving a request for instructions to perform the one or more operations on the one or more fields of the data structure, the instructions being in accordance with a second syntax; generating a translation of the definition of the one or more operations from the first syntax to the second syntax; generating instructions for accessing the one or more fields of the data structure; and generating the instructions in accordance with the second syntax for performing the one or more operations on the one or more fields of the data structure.

Generating a translation of operations for a data structure

A computer-implemented method includes receiving a request to populate one or more first fields of a first data structure with one or more first values that are at least partly based on one or more second values in one or more second fields of a second data structure; generating, based on the request, one or more instructions to (i) select the one or more second values, (ii) generate the one or more first values, and (iii) populate the one or more first fields of the first data structure with the one or more first values; and sending the one or more instructions to a computing device for insertion into a computer program that is executable by the computing device.

Transferring values among fields in data structures

A method performed by one or more processing devices includes receiving a request to generate a computer program; generating, in response to receiving the request, a first user interface that when rendered on a display device renders: one or more first visual representations of one or more first fields of the first resource; and one or more second visual representations of one or more second fields of the second resource; receiving, through the first user interface, a selection of: at least one of the one or more first visual representations; and at least one of the one or more second visual representations; receiving, through a second user interface, data specifying the one or more operations; and updating, based on the received data, the computer program with one or more third instructions to perform the one or more operations.

Generating a user interface for sending data among resources

A computer-implemented method includes receiving a request to search for a data structure associated with one or more search criteria; searching, an index in a data repository, for the data structure, with the search based on the one or more search criteria; identifying, based on searching, the data structure associated with the one or more search criteria; obtaining a definition for the data structure; generating an interface that when presented on a computing device renders a representation of the definition for the data structure; and sending the interface to the computing device associated with the request to search.

Growing community of inventors

San Jose, CA, United States of America

Jianjun Wen