International Business Machines Corporation

Globalfoundries Inc.

Victor V Zyuban

Pradip Bose

Jude A Rivers

Alper Buyuktosunoglu

Hans Mikael Jacobson

Jeonghee Shin

Karthick Rajamani

Charles Robert Lefurgy

Malcolm Scott Allen-Ware

Michael K Gschwind

Sumedh Wasudeo Sathaye

Gregory Scott Still

Todd Jon Rosedahl

John-David Wellman

Erik Richter Altman

Guillermo Jesus Silva

Zhigang Hu

Balaram Sinharoy

Kirk David Peterson

Brian William Thompto

Eren Kursun

Michael Stephen Floyd

Vijayalakshmi Srinivasan

Srinivasan Ramani

Prabhakar Nandavar Kudva

Joshua D Friedrich

Philip George Emma

David James Hathaway

Robert K Montoye

Jon Alfred Casey

Jeffrey Haskell Derby

Mayan Moudgill

Thomas Anthony Wassick

Stanley Everett Schuster

Sameh W Asaad

Alan J Drake

Chen-Yong Cher

David Meltzer

Sungjun Chun

Fredy Daniel Neeser

Peter William Cook

Erdem Hokenek

Christopher J Gonzalez

Moinuddin K Qureshi

Jaime Humberto Moreno

Shay Ben-David

Ayal Zaks

Uzi Shvadron

Nagashyamala R Dhanwada

Bjorn P Christensen

Jamie H Moreno

Daniel M Citron

Eko Lisuwandi

A method and apparatus for adjusting a frequency of a processor is disclosed herein. In one embodiment, the method includes inhibiting one or more processor cores from exiting an idle state. The method further includes determining a number of processor cores requesting exit from the idle state and a number of non-idle processor cores. The method also includes selecting a maximum frequency for the inhibited processor cores based on the number of inhibited processor cores requesting exit from the idle state and the number of non-idle processor cores. The method includes setting the maximum frequency for the inhibited processor cores, and then uninhibiting the processor cores requesting exit from the idle state.

Management of core power state transition in a microprocessor

The embodiments described herein relate to dynamically detecting a frequency change condition for microprocessor performance. An instruction is received, and a frequency change condition associated with the received instruction is dynamically detected. A frequency modulation is performed in response to the dynamic detection. The frequency modulation includes selecting a second frequency for optimal instruction processing different from a first frequency, the first frequency being a default operating frequency of the microprocessor. Execution of the instruction is completed at the second frequency. Accordingly, incoming execution instructions are logically analyzed, and the processor frequency is selectively modified based on associated instruction conditions.

Dynamic workload frequency optimization

A system for adjusting a frequency of a processor is disclosed herein. The system includes a processor and a memory, where the memory includes a program configured to adjust a frequency of a multi-core processor. The operations include determining a total current and a temperature of the multi-core processor and estimating a leakage current for the multi-core processor. The operations also include calculating a switching current by subtracting the leakage current from the total current and calculating an effective switching capacitance based at least in part on the switching current. The operations also include calculating a workload activity factor by dividing the effective switching capacitance by a predetermined effective switching capacitance stored in vital product data, and enforcing a turbo frequency limit of the multi-core processor based on the workload activity factor.

Deterministic current based frequency optimization of processor chip

A method for adjusting a frequency of a processor is disclosed herein. In one embodiment, the method includes inhibiting one or more processor cores from exiting an idle state. The method further includes determining a number of processor cores requesting exit from the idle state and a number of non-idle processor cores. The method also includes selecting a maximum frequency for the inhibited and non-idle processor cores based on the number of inhibited processor cores requesting exit from the idle state and the number of non-idle processor cores. The method includes setting the maximum frequency for both the inhibited and the non-idle processor cores, and then uninhibiting the processor cores requesting exit from the idle state.

A method for adjusting a frequency of a processor is disclosed herein. In one embodiment, the method includes determining a total current and a temperature of the multi-core processor and estimating a leakage current for the multi-core processor. The method also includes calculating a switching current by subtracting the leakage current from the total current. The method also includes calculating an effective switching capacitance based at least in part on the switching current. The method also includes calculating a workload activity factor by dividing the effective switching capacitance by a predetermined effective switching capacitance stored in vital product data, and enforcing a turbo frequency limit of the multi-core processor based on the workload activity factor.

A system for adjusting a frequency of a processor is disclosed herein. The system includes a processor and a memory. The memory stores program code, which, when executed on the processor, performs an operation for adjusting a frequency of a processor. The operation includes inhibiting one or more processor cores from exiting an idle state. The operation further includes determining a number of processor cores requesting exit from the idle state and a number of non-idle processor cores. The operation also includes selecting a maximum frequency for the inhibited and non-idle processor cores based on the number of inhibited processor cores requesting exit from the idle state and the number of non-idle processor cores. The operation includes setting the maximum frequency for both the inhibited and the non-idle processor cores, and then uninhibiting the processor cores requesting exit from the idle state.

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 an integrated circuit with power gating. A power header switch is configured to connect and disconnect a circuit to a common voltage source. The circuit is powered off circuit when disconnected. A multiplexer selectably connects a charge pump or common voltage source to a gate terminal of the power header switch. The charge pump provides a higher voltage to the gate terminal than the common voltage source. A controller is configured to control a selection of the multiplexer to the charge pump and the common voltage source. The controller is configured to disconnect the charge pump from the gate terminal and connect the common voltage source to the gate terminal of the power header switch in response to conditions: a prediction of a demand core power-up request, an increase in a gate leakage current, and/or a reduction in temperature of the powered off circuit.

Predictively turning off a charge pump supplying voltage for overdriving gates of the power switch header in a microprocessor with power gating

A mechanism is provided for an integrated circuit with power gating. A power switch is configured to connect and disconnect circuits to a common voltage source. A capacitor tank is configured to supply wakeup charge to a given circuit. A controllable element is connected to the given circuit and to the capacitor tank. The controllable element is configured to controllably connect and disconnect the capacitor tank to the given circuit in order to supply the wakeup charge to the given circuit. The controllable element is configured to, responsive to the power switch disconnecting the given circuit from the common voltage source and to the given circuit being turned on to wakeup, supply the wakeup charge to the given circuit being turned on by transferring the wakeup charge from the capacitor tank to the given circuit. This reduces the electrical charge transferred from the circuits connected to the common voltage source.

Accelerating microprocessor core wake up via charge from capacitance tank without introducing noise on power grid of running microprocessor cores

A method for generating a power model for a device includes identifying a device-level set of power contributors for a given state of the device, wherein each power contributor in the device-level set of power contributors contributes to power dissipation when the device is in the given state, and generating the power model for the device based on the device-level set of power contributors, wherein the power model is independent of process, voltage, and temperature.

Contributor-based power modeling of microprocessor components

According to one embodiment, a system is provided that includes at least one power gated component and two or more power switch transistors configured to provide one or more conductive paths between a common power supply rail, the at least one power gated component, and a ground. The two or more power switch transistors each include a source terminal, a drain terminal, and a gate terminal configured to control current flow between the source and drain terminals. The system also includes a rotating voltage control coupled to the gate terminals and configured to apply a sequence of control signals rotating between an on-state and an off-state to each of the gate terminals while the at least one power gated component is turned on. A switch activation ratio level is programmable to set a number of power switch transistors in the on-state relative to a total number of power switch transistors.

Rotating voltage control

A mechanism is provided for an integrated circuit with power gating. A power header switch is configured to connect and disconnect any one of multiple circuits to a common voltage source, where a powered off circuit is disconnected from the common voltage source. A power-up sequencer includes an initial stages power-up component and a final stages power-up component. The final stages power-up component is configured to execute final stages of a power-up process for the powered off circuit, and the initial stages power-up component is configured to execute initial stages of the power-up process for the powered off circuit. The initial stages power-up component is activated in response to a predictive power-up request.

Accelerating the microprocessor core wakeup by predictively executing a subset of the power-up sequence

In one aspect, a method of enhancing semiconductor chip process variability and lifetime reliability through a three-dimensional (3D) integration applied to electronic packaging is disclosed. Also provided is an arrangement for implementing the inventive method. In another aspect, a method and on-chip controller are disclosed for enhancing semiconductor chip process variability and lifetime reliability through a three-dimensional (3D) integration applied to electronic packaging. Also provided is an on-chip reliability/variability controller arrangement for implementing the inventive method. In yet another aspect, base semiconductor chips, each comprising a plurality of chiplets, are manufactured and tested. For a base semiconductor chip having at least one non-functional chiplet, at least one repair semiconductor chiplet is vertically stacked. A functional multi-chip assembly is formed, which provides the same functionality as a base semiconductor chip in which all chiplets are functional.

Semiconductor chip repair by stacking of a base semiconductor chip and a repair semiconductor chip

Growing community of inventors

Peekskill, NY, United States of America

Victor V Zyuban