Amazon Technologies, Inc.

Bradley R Bebee

Bryan Thompson

Thomas James Lewis

Olaf Hartig

Divyakala Vel

Sainath Chowdary Mallidi

Karthik Gurumoorthy Subramanya Bharathy

George Karypis

Serkan Turgut

Ilie Gabriel Tanase

Ankesh Khandelwal

Herman Yusef Polloni Giacaman

Yigit Kiran

Adesoji Adeshina

Navtanay Sinha

Xiang Song

Michael Schmidt

Geo T Varkey

Ankit Gupta

Tiago Lima Salmito

Systems and methods are described for dynamically allocating memory for responding to requests. In some aspects, a thread may be obtained, where the thread includes a request to use a portion of system memory to perform a query of a database, such as a graph database. The system may determine that the portion of memory is greater than available memory of a maximum managed memory size of the system, and may block the thread until additional memory becomes available. The maximum managed memory size may be configurable, and may be less than a total memory of the system. The system may then determine, based at least in part on a change in the available system memory, that the available system memory is equal to or greater than the portion of memory to satisfy the request, and cause the system to process the thread to provide a response to the query.

Dynamic bounded memory allocation

A computer-implemented method performed at least in part by a graph database configured to store at least one graph dataset. The method includes receiving a graph query configured to be performed against a machine learning model, and communicating the graph query with a machine learning system that is configured to use the machine learning model to obtain model inference results and communicate those model inference results to the graph database application. The graph database provides query results based at least in part on the model inference results to an entity.

Using graph queries to obtain results from machine learning models

Source code of a graph analysis program expressed in a platform-independent language which supports linear algebra primitives is obtained. An executable version of the program is generated, which includes an invocation of a function of a parallel programming library optimized for a particular hardware platform. A result of executing the program is stored.

Acceleration techniques for graph analysis programs

A bit vector representing access permissions associated with respective vertices of a graph data set is generated. At least a portion of the bit vector is read, and a first graph analytics algorithm is performed. The algorithm comprises determining, based at least in part on a portion of the bit vector, whether access permission to one or more vertices of the graph data set is granted.

Security management for graph analytics

Growing community of inventors

Washington, DC, United States of America

Bradley R Bebee