Advanced Micro Devices Corporation

Michael L Golden

Richard Martin Born

Gokul Subramani Ramalingam Lakshmi Devi

William Louie Walker

Marius Evers

Amitabh Mehra

Xiaojie He

Sriram Srinivasan

Richard E Klass

Sneha Komatireddy

Xiuting Kaleen C Man

Samuel D Naffziger

Stephen Victor Kosonocky

Larry David Hewitt

Russell J Schreiber

Hamid Partovi

Edward J McLellan

Michael Estlick

Paul Blinzer

Keith Allen Kasprak

Eric W Busta

Magiting M Talisayon

James A Wingfield

Srikanth Masanam

Seng Oon Toh

John Yong

Upasanah Swaminathan

Sean M O′Mullan

Ripal Butani

Methods and systems for facilitating improved power consumption control of a plurality of processing cores are disclosed. The methods improve the power consumption control by performing power throttling based on a determined excess power consumption. The methods include the steps of: monitoring using at least one event count component in the respective processing core a plurality of distributed events; calculating an accumulated weighted sum of the distributed events from the event count component; determining an excess power consumption by comparing the accumulated weighted sum with a threshold power value; and adjusting power consumption of the respective processing core based on the determined excess power consumption.

System and method for controlling power consumption in processor using interconnected event counters and weighted sum accumulators

Within a processing system, thread count asymmetries manifest when one or more cores of a processing device are disabled. To determine such thread count asymmetries, discovery operations are performed to determine thread count asymmetries for one or more hierarchy levels of a processing device based on a number of threads per enumerated instance within the hierarchy level. In response to the determining a thread count asymmetry, one thread identifier for each enumerated instance within the asymmetric hierarchy level is defined to determine a representation of the asymmetry. Using the representation of the symmetry, software tasks associated with one or more application within the processing system are performed.

Hierarchical asymmetric core attribute detection

An integrated circuit includes one or more processing units that execute instructions that employ a register file, control logic creates a pre-startup register free list, prior to normal operation of at least one of the processing units, that includes a list of registers devoid of defective registers. In some implementations, no column and row repair information is provided to register file repair logic. In certain examples, the register file is configured as a repair-less register file. During normal operation of the one or more processing units, the integrated circuit employs the pre-startup register free list to select registers in a register file for the executing instructions. Associated methods are also presented.

Methods and apparatus for managing register free lists

Methods and apparatuses control electrical current supplied to a plurality of processing units in a multi-processor system. A plurality of current usage information corresponding to the processing units are received by a controller to determine a threshold current for each of the processing units. The controller determines a frequency reduction action and an instructions-per-cycle (IPC) reduction action for the each of the processing units based on the threshold current and regulates operations of the processing units based on the determined frequency and IPC reduction actions.

System and method for controlling electrical current supply in a multi-processor core system via instruction per cycle reduction

A processor core associated with a first cache initiates entry into a powered-down state. In response, information representing a set of entries of the first cache are stored in a retention region that receives a retention voltage while the processor core is in a powered-down state. Information indicating one or more invalidated entries of the set of entries is also stored in the retention region. In response to the processor core initiating exit from the powered-down state, entries of the first cache are restored using the stored information representing the entries and the stored information indicating the at least one invalidated entry.

Retaining cache entries of a processor core during a powered-down state

A processor employs a set of replica paths at a processor to determine an operating frequency and voltage for the processor. The replica paths each represent one or more circuit paths at a functional module of the processor. The delays at the replica paths are normalized to increase the likelihood that the replica paths accurately represent the behavior of the circuit paths of the functional module. After normalization, a distribution of delay values is generated by varying, at each replica path, the delay at an output node of the replica path until a mismatch is detected between a signal at the output node of the replica path and the delayed representation of the signal. The resulting distribution of delay values can then be adjusted based on variations in reference voltages at the replica paths to account for potential distribution errors resulting from the reference voltage variations.

Replica path timing adjustment and normalization for adaptive voltage and frequency scaling

In one embodiment, a processing node includes a first processor core and a second processor core. The first processor core includes a first cache memory, such as an L2 cache, for example. The second processor core includes a second cache memory, such as an L2 cache memory. The processing node further includes a configuration unit that is coupled to the first processor core and the second processor core. The configuration unit may selectably disable portions of the first and the second cache memories.

Reconfigurable processing node including first and second processor cores

In one embodiment, a processing node includes a plurality of processor cores each including a cache memory coupled to a cache monitor unit and to a configuration unit. Each cache monitor unit may be configured to independently monitor a current utilization of the cache memory to which it is coupled and to determine whether the current utilization is below a predetermined utilization value. The configuration unit may selectably disable one or more portions of the cache memory in response to the cache monitor unit determining that the current utilization is below the predetermined utilization value.

Dynamic reconfiguration of cache memory

A flip-flop circuit includes a differential stage coupled to a transparent latch. Respective sides of the differential stage, referred to as the &ldquo;output side&rdquo; and the &ldquo;reference side,&rdquo; are precharged high during a precharge phase. During an evaluation phase, the state of a data input signal is sensed. Depending upon the state of the data input signal, either the output side or the reference side is discharged. Also, during the evaluation phase, the transparent latch is enabled, and thereby samples and stores an output signal from the output side of the differential stage. Upon initiation of the next precharge phase, the transparent latch is quickly disabled (i.e., is placed in an opaque state), and retains its present state. Since only a single side of the differential stage is used to drive the transparent latch, the differential stage may advantageously be implemented in an asymmetric fashion. In yet an additional embodiment, complex logic may be added to the differential stage of the flip-flop circuit.

Low-power flip-flop circuit employing an asymmetric differential stage

Growing community of inventors

Santa Clara, CA, United States of America

Michael L Golden