Microsoft Technology Licensing, LLC

Siddhartha Puri

Samuel James McKelvie

Raymond Edward Ozzie

Christopher G Kaler

David Maxwell Chickering

Erik B Christensen

Kenneth David Wolf

Bradley Gene Calder

Denis X Charles

John D Mehr

Elliot Lee Waingold

Michael Steven Vernal

Matt Pope

Sven Seuken

Stephen James Millet

David Wortendyke

Giovanni Moises Della-Libera

James Ernest Johnson

Tirunelveli R Vishwanath

Lance Olson

Richard Douglas Hill

Umesh Madan

Stefan R Batres

Jeffrey C Schlimmer

Michael James Coulson

Stefan Harrington Pharies

John David Doty

Craig Andrew Critchley

Michael Marucheck

Vijay Gajjala

Krishnan Srinivasan

Uday Sooryakant Hegde

Tomasz Janczuk

Geary L Eppley

Matthew Thomas Tavis

Stephen T Swartz

David Edwin Levin

Shy Cohen

Martin James Gudgin

Max A Feingold

Richard Dievendorff

Douglas M Purdy

Natasha Harish Jethanandani

David Owen Driver

Saurab Nog

Aaron Stern

Douglas A Walter

Jesus Ruiz-Scougall

Anastasios Kasiolas

Kapil Gupta

Vipul A Modi

Ryan Thomas Sturgell

Mohammad Makarechian

Alexander Martin DeJarnatt

Evgeny Osovetsky

Sowmyanarayanan K Srinivasan

Scott Christopher Seely

Jonathan T Wheeler

Erin Honeycutt

Yaniv Pessach

Mauro Ottaviani

Chaitanya Dutt Upadhyay

Michael Thomas Dice

Eugene Shvets

Yasser Shohoud

Alfred M Lee, Iv

Jaydeep P Marathe

Anand Rjagopalan

Mohamed-Hany Essam Ramadan

Serge Sverdlov

Vaithialingam B Balayogan

Embodiments of erasure coded storage aggregation are disclosed. The erasure coded storage aggregation includes storing a data file as erasure coded fragments in a plurality of nodes of one or more data centers. The erasure coded storage aggregation further includes monitoring an access frequency of the data file. Based on the comparison between the access frequency and a predetermined threshold, the data file is either reconstructed from the erasure coded fragments and stored in a storage node or retained as erasure coded fragments in the plurality of nodes of the one or more data centers.

Erasure coded storage aggregation in data centers

Described is a technology by which a series of loop nests corresponding to source code are detected by a compiler, with the series of loop nests tiled together, (thereby increasing the ratio of cache hits to misses in a multi-processor environment). The compiler transforms the series of loop nests into a plurality of tile loops within a controller loop, including using dependency analysis to determine which results from a tile loop need to be pre-computed before another tile loop. For dependency analysis, the compiler may use a directed acyclic graph as a high-level intermediate representation, and split the graph into sub-graphs each representing an array. The compiler uses descriptors processed from the graph to determine the controller loop and the tile loops within that controller loop.

Tiling across loop nests with possible recomputation

Providing a market design for a peer-to-peer resource exchange system. Prices for a plurality of resources such as storage space, upload bandwidth, and download bandwidth are calculated and balanced based on previous resource prices, a supply of the resources, and a demand for the resources. Further, prices for operations such as storage and retrieval are determined such that a total of the payments to resource suppliers equals a total of the payments received from the resource consumers. In some embodiments, incoming data operation requests are allocated to the peers such that equilibrium among the peers is achieved.

Market design for a resource exchange system

A mechanism for sending structured data using a corresponding byte stream. Upon accessing structured data such as a Simple Object Access Protocol (SOAP) envelope, a byte stream is generated. The byte stream includes bytes that represent the structured data, as well as a collection of bytes that represents properties about the byte stream such as, for example, a mode of communication. The byte stream may then be passed to a communication module (e.g., a TCP or Named Pipes module) capable of receiving and transmitting the byte stream.

Mechanism for binding a structured data protocol to a protocol offering up byte streams

A convolutional neural network is implemented on a graphics processing unit. The network is then trained through a series of forward and backward passes, with convolutional kernels and bias matrices modified on each backward pass according to a gradient of an error function. The implementation takes advantage of parallel processing capabilities of pixel shader units on a GPU, and utilizes a set of start-to-finish formulas to program the computations on the pixel shaders. Input and output to the program is done through textures, and a multi-pass summation process is used when sums are needed across pixel shader unit registers.

Training convolutional neural networks on graphics processing units

A web services namespace pertains to an infrastructure for enabling creation of a wide variety of applications. The infrastructure provides a foundation for building message-based applications of various scale and complexity. The infrastructure or framework provides APIs for basic messaging, secure messaging, reliable messaging and transacted messaging. In some embodiments, the associated APIs are factored into a hierarchy of namespaces in a manner that balances utility, usability, extensibility and versionability.

Interface infrastructure for creating and interacting with web services

Growing community of inventors

Sammamish, WA, United States of America

Siddhartha Puri