International Business Machines Corporation

Taiwan Semiconductor Manufacturing Comp. Ltd.

Globalfoundries Inc.

Alan J Drake

Michael Stephen Floyd

Charles Robert Lefurgy

Malcolm Scott Allen-Ware

Karthick Rajamani

Gary Dale Carpenter

Phillip John Restle

Fadi Hikmat Gebara

Xiaobin Yuan

Pawel Owczarczyk

Marshall Dale Tiner

Thomas J Bucelot

Jason D Hibbeler

William Robert Reohr

Liang-Teck Pang

Michael G R Thomson

Robert Allen Groves

Guillermo Jesus Silva

Todd Jon Rosedahl

Andrew Kenneth Martin

Heather Lynn Hanson

Robert Matthew Senger

Timothy Glen Hallett

Jordan Ross Keuseman

Aj KleinOsowski

Yong I Kim

Jente Benedict Kuang

Gregory Scott Still

Jeremy Daniel Schaub

David T Hui

John Patrick Keane

Eren Kursun

Edmund Juris Sprogis

Alan S Roth

Hung Cai Ngo

Eric Soenen

Wei T Huang

Barry P Linder

Michael Jay Shapiro

Tilman Gloekler

Aditya Bansal

Ronald Jay Bolam

Huajun Wen

Rachel Gordin

Steven W Mittl

Jerry C Kao

Francesco A Campisano

Harmander Singh

Mrigank Sharad

David W Shan

Rajiv V Joshi

Malcolm Scott Ware

Kevin John Nowka

Jon Alfred Casey

Thomas Anthony Wassick

Uttam Shymalindu Ghoshal

John Bruce Carter

Ethan Harrison Cannon

Tobias Webel

Sungjun Chun

Victor V Zyuban

Bishop Chapman Brock

Robert Walter Berry, Jr

Norman Karl James

Chandler Todd McDowell

Jeonghee Shin

Jeffrey L Burns

Russell Kinder

Joshua D Friedrich

Ruopeng Wang

Harmander Singh Deogun

Richard L Willaman

Malcom S Allen-Ware

Aj Klein Osowski

A distributed power management system is configured determine a node power consumption of a node during a first time interval. The system can determine a node power cap. The system can determine a proportional component power budget for a component of the node based, at least in part, on the node power consumption and a component power consumption. The system can determine a power budget for the component for a second time interval based, at least in part on the proportional component power budget.

Distributed power budgeting

An approach is provided in which a multi-core processor's first core determines whether it controls a system frequency that drives a group of cores included in the multi-core processor. When the first core is not controlling the system frequency for the group of cores, the first core uses an internal voltage control module to provide control information to the first core's programmable voltage regulator and, in turn, independently control the first core's voltage level. When the first core is controlling the system frequency, the first core receives voltage control information from pervasive control to control the first core's voltage levels.

Multi-core dynamic frequency control system

A multi-phase buck converter comprises a first comparator, a second comparator and a counter. The first comparator has a first node connected to a first voltage reference and a second node. The second comparator has a first node connected to a second voltage reference and a second node. The second node of the second comparator and the second node of the first comparator are together connected to an input voltage from an active phase of the buck converter. The counter is configured to adjust a number of active phases of the buck converter based on the input voltage.

Multiphase buck converter and method for operating the same

The present disclosure relates to voltage regulation techniques. In some embodiments, a voltage regulator is configured to regulate an output voltage based on a reference voltage. The voltage regulator comprises an analog-to-digital converter, an encoder, a decoder and a power stage. The analog-to-digital converter receives the reference voltage and an output voltage of the voltage regulator and provides a digital error signal. The encoder is coupled to the analog-to-digital converter output and configured to provide a multi-bit digital control signal based upon a present value of the digital error signal, a plurality of pre-determined coefficients, and a plurality of previous values of the digital error signal. The decoder is coupled to the encoder and configured to generate a plurality of control signals based on the multi-bit digital control signal. The power stage comprises a plurality of power cells which are coupled to a power supply and which receive the plurality of control signals, respectively.

Smart voltage regulation techniques

A method for managing a processor, the processor comprising a common supply rail and processor cores being connected to the common supply rail, wherein each processor core comprises a core unit, wherein the method comprises detecting idle state exits indicated by the core units; and delaying a command execution of at least one of the core units indicating an idle state exit when the number of idle state exits exceeds a predetermined threshold idle state exit number may reduce voltage droops due to several processor cores leaving the idle state at the same time.

System and method for controlling idle state exits to manage DI/DT issues

An system and method are configured to degrade a memory cell PFET voltage based on a sensor reading of a current operating point. This will enable additional control over the SRAM device, particularly during a write operation. In one embodiment, a system of SRAM memory devices is configured as a smart sensor with real-time corrective circuit action. The system and method samples write and read timing operations and is adaptable by performing real-time corrective action. The degrading of PFET voltage to reduce it strength and improve write characteristics include an implementation that includes a charge pump controllable for altering by decreasing a voltage applied to the PFET of a selected memory cell. In a further embodiment, an edge detector is built into the circuit that real-time assesses the strength of the memory write operation. In a further implementation, control logic functions as a Finite State Machine.

Mitigation scheme for SRAM functionality

A wide bandwidth resonant clock distribution comprises a clock grid configured to distribute a clock signal to a plurality of components of an integrated circuit and a tunable sector buffer configured to receive the clock signal and provide an output to the clock grid. The tunable sector buffer is configured to set latency and slew rate of the clock signal based on an identified resonant or non-resonant mode.

Tunable sector buffer for wide bandwidth resonant global clock distribution

A clock driver is provided. The clock driver includes a multi-stage delay cell having an input, a positive pulse driving branch, a negative pulse driving branch, and an output. The input is for receiving an original version of a reference clock signal input to the clock driver and used to generate a global clock signal. The output is connected to the positive pulse driving branch and the negative pulse driving branch. The clock driver further includes a pulse generator having positive and negative pulse generator portions respectively connected to outputs of the positive and negative pulse driving branches. The pulse generator generates, at any given time, one of a positive pulse and a negative pulse responsive to a positive pulse enable signal and a negative pulse enable signal, respectively, and the original version of the reference clock signal input to the clock driver without modification.

Clock buffers with pulse drive capability for power efficiency

A mechanism is provided for determining a modeled age of a set of interconnect groups in a set of cores in a set of multi-core processors. For each interconnect group in the set of interconnect groups in the set of cores on the set of multi-core processors, a determination is made of a current modeled age of the interconnect group. A determination is then made as to whether at least one current modeled age of the interconnect group for the set of interconnect groups is greater than an end-of-life value. Responsive to at least one current modeled age of the interconnect group being greater than the end-of-life value, an indication to take corrective action with the at least one associated interconnect group is sent.

Managing interconnect electromigration effects

A mechanism is provided for implementing an operational parameter change within the data processing system based on an identified degradation. One or more degradations existing in the data processing system are identified based on a set of degradation values obtained from a set of degradation sensors. A determination is made as to whether one or more operational parameters need to be modified based on the one or more identified degradations. Responsive to determining that the one or more operational parameters need to be modified based on the one or more identified degradations, an input change is implemented to a one or more control devices in order that the one or more operational parameters are modified.

Dynamic adjustment of operational parameters to compensate for sensor based measurements of circuit degradation

A computing circuit that includes clocked circuitry, a controller, and a clock generator. The clocked circuitry is configured to receive data and to perform data manipulation on the data based on a first clock signal. The controller is configured to control the transmission of the data to the clocked circuitry. The clock generator is configured to receive as inputs a second clock signal and a delay control signal from the controller, and to delay the second clock signal to generate the first clock signal. The clock generator includes a main delay component configured to receive the second clock signal and to output the first clock signal. The clock generator also includes a switchable delay component connected in parallel with the main delay component, where the switchable delay component is configured to receive as an input the delay control signal from the controller.

Programmable delay circuit

A critical path monitor (CPM) having a set of split paths is configured in an integrated circuit (IC) that includes a corresponding set of critical paths. A first and a second split path is configured with a first and a second simulated delay sections and fine delay sections, respectively. A delay of each of the first and second fine delay sections is adjustable in several steps. The delay of the first fine delay section is adjustable differently from the delay of the second fine delay section in response to a common operating condition change. Differently adjusting the delays of the first and the second fine delay sections causes an edge of a pulse to be synchronized between a first edge detector located after the first simulated delay section and a second edge detector located after the second simulated delay section.

Adjustable delay calibration in a critical path monitor

Growing community of inventors

Round Rock, TX, United States of America

Alan J Drake