Mihai Dan Budiu

8504874, Aug 6, 2013

Sep 21, 2010

12/886566

Moises Goldszmidt - Palo Alto CA, US
Mihai Budiu - Sunnyvale CA, US
Yue Zhang - Sunnyvale CA, US
Michael Pechuk - Sammamish WA, US

Microsoft Corporation - Redmond WA

G06F 11/00

714 25

In a distributed system a plurality of devices (including computing units, storage and communication units) are monitored by an automated repair service that uses sensors and performs one or more repair actions on computing devices that are found to fail according to repair policies. The repair actions include automated repair actions and non-automated repair actions. The health of the computing devices is recorded in the form of states along with the repair actions that were performed on the computing devices and the times at which the repair actions were performed, and events generated by both sensors and the devices themselves. After some period of the time, the history of states of each device, the events, and the repair actions performed on the computing devices are analyzed to determine the effectiveness of the repair actions. A statistical analysis is performed based on the cost of each repair action and the determined effectiveness of each repair action, and one or more of the policies may be adjusted, as well as determining from the signals and events from the sensors whether the sensors themselves require adjustment.

20100180240, Jul 15, 2010

Jan 13, 2009

12/353168

John D. Davis - San Francisco CA, US
Mihai Budiu - Sunnyvale CA, US
Hari Kannan - Stanford CA, US

G06F 17/50

716 5

The Global Dynamic Critical Path is used to optimize the design of a system-on-a-chip (SoC), where hardware modules are in different clock domains. Control signal transitions of the hardware modules are analyzed to identify the Global Dynamic Critical Path. Rules are provided for handling specific situations such as when concurrent input control signals are received by a hardware module. A configuration of the hardware modules is modified in successive iterations to converge at an optimum design, based on a cost function. The cost function can account for processing time as well as other metrics, such as power consumed. For example, during the iterations, hardware modules which are in the Global Dynamic Critical Path can have their clock speed increased and/or additional resources can be added, while hardware modules which are not in the Global Dynamic Critical Path can have their clock speed decreased and/or unnecessary resources can be removed.

20100205588, Aug 12, 2010

Feb 9, 2009

12/368231

Yuan Yu - Cupertino CA, US
Dennis Fetterly - Belmont CA, US
Michael Isard - San Francisco CA, US
Ulfar Erlingsson - Reykjavik, IS
Mihai Budiu - Sunnyvale CA, US

MICROSOFT CORPORATION - Redmond WA

G06F 9/45
G06F 15/16

717149, 709203, 717153

General-purpose distributed data-parallel computing using a high-level language is disclosed. Data parallel portions of a sequential program that is written by a developer in a high-level language are automatically translated into a distributed execution plan. The distributed execution plan is then executed on large compute clusters. Thus, the developer is allowed to write the program using familiar programming constructs in the high level language. Moreover, developers without experience with distributed compute systems are able to take advantage of such systems.

20110199380, Aug 18, 2011

Feb 15, 2010

12/705633

Mihai Budiu - Sunnyvale CA, US

Microsoft Corporation - Redmond WA

G06T 11/20
G06F 3/048

345440, 715769, 715810

A system for visualizing correlations between attributes in a data set or across multiple data sets is provided. A user may view a graphical representation (e.g., a histogram) of attribute values for a first attribute. The user may assign a variety of graphical indicators to various value ranges of the first attribute. The user may view a graphical representation of the second attributes. The user may “drag and drop” the graphical representation of the first attributes onto the graphical representation of the second attributes. The graphical representation of the second attributes may be updated to incorporate the graphical elements assigned by the user to the value ranges of the first attribute. The user may visually see potential correlations between the first and the second attributes based on the graphical elements associated with the first attributes displayed with the associated second attributes.

20120254597, Oct 4, 2012

Mar 29, 2011

13/074006

Daniel Delling - Mountain View CA, US
Mihai Budiu - Sunnyvale CA, US
Renato F. Werneck - San Francisco CA, US

Microsoft Corporation - Redmond WA

G06F 9/38

712233, 712E09045

A distributed data-parallel execution (DDPE) system splits a computational problem into a plurality of sub-problems using a branch-and-bound algorithm, designates a synchronous stop time for a “plurality of processors” (for example, a cluster) for each round of execution, processes the search tree by recursively using a branch-and-bound algorithm in multiple rounds (without inter-processor communications), determines if further processing is required based on the processing round state data, and terminates processing on the processors when processing is completed.

20130125099, May 16, 2013

Nov 14, 2011

13/295107

Mihai Budiu - Sunnyvale CA, US
Gordon D. Plotkin - Edinburgh, GB
Joel Galenson - Berkeley CA, US

Microsoft Corporation - Redmond WA

G06F 9/45

717140

A modular compiler architecture utilizes partial compiler modules that cooperatively produce object code for operation on a complex execution infrastructure. The partial compilers may invoke the services of other partial compilers, wherein each partial compiler operates as a self-contained “black-box” module. This structure, in turn, may allow the partial compilers of such implementations to be arranged in modular hierarchies for multi-level compilation and specialization of each partial compiler. These various implementations, in turn, produce compiled programs able to correctly run on large computer clusters comprising a mix of computational resources (machines, multiple cores, graphics cards, SQL server engines, etc.). Certain implementations may also be directed to compilers comprising modular partial compilers, and partial compilers may be formed from generalized forms of traditional compilers. Further disclosed is a set of high-level operations that manipulate partial compilers.

20170371721, Dec 28, 2017

Jul 10, 2017

15/645620

- Redmond WA, US
Dennis Fetterly - Belmont CA, US
Michael Isard - San Francisco CA, US
Ulfar Erlingsson - Palo Alto CA, US
Mihai Budiu - Sunnyvale CA, US

Microsoft Technology Licensing, LLC - Redmond WA

G06F 9/52
G06F 9/45
G06F 9/44

General-purpose distributed data-parallel computing using a high-level language is disclosed. Data parallel portions of a sequential program that is written by a developer in a high-level language are automatically translated into a distributed execution plan. The distributed execution plan is then executed on large compute clusters. Thus, the developer is allowed to write the program using familiar programming constructs in the high level language. Moreover, developers without experience with distributed compute systems are able to take advantage of such systems.

20170093987, Mar 30, 2017

Sep 24, 2015

14/864088

- Palo Alto CA, US
Mihai Budiu - Sunnyvale CA, US
Changhoon Kim - Palo Alto CA, US

H04L 29/08

A synchronous packet-processing pipeline whose data paths are populated with data-plane stateful processing units (DSPUs) is provided. A DSPU is a programmable processor whose operations are synchronous with the dataflow of the packet-processing pipeline. A DSPU performs every computation with fixed latency. Each DSPU is capable of maintaining a set of states and perform its computations based on its maintained set of states. The programming of a DSPU determines how and when the DSPU updates one of its maintained states. Such programming may configure the DSPU to update the state based on its received packet data, or to change the state regardless of the received packet data.