Abstract: Allocating power between multiple central processing units (CPUs) in a multi-CPU processor based on total current availability and individual CPU quality-of-service (QoS) requirements is disclosed. Current from a power rail is allocated to CPUs by a global current manger (GCM) circuit related to performance criteria set by CPUs. The CPUs can request increased current allocation from the GCM circuit, such as in response to executing a higher performance task. If the increased current allocation request keeps total current on the power rail within its maximum rail current limit, the GCM circuit approves the request to allow the CPU increased current allocation. This can allow CPUs executing higher performance tasks to have a larger current allocation than CPUs executing lower performance tasks without the maximum rail current limit being exceeded, and without having to necessarily lower voltage of the power rail, which could unnecessarily lower performance of all CPUs.

Abstract: Adaptive voltage modulation circuits for adjusting supply voltage to reduce supply voltage droops and minimize power consumption are provided. In one aspect, an adaptive voltage modulation circuit detects a supply voltage droop by detecting when a supply voltage falls below a droop threshold voltage, and adjusts a clock signal provided to a load circuit in response to a supply voltage droop. The adaptive voltage modulation circuit keeps a count of the number of clock signal cycles during which the supply voltage is below the droop threshold voltage. The adaptive voltage modulation circuit increases the supply voltage in response to the count exceeding an upper threshold value, and decreases the supply voltage in response to the count being less than a lower threshold value at an end of a defined period. The adaptive voltage modulation circuit can reduce the time a load circuit operates with reduced frequency while minimizing power consumption.

Abstract: The disclosure generally relates to dynamic clock and voltage scaling (DCVS) based on program phase. For example, during each program phase, a first hardware counter may count each cycle where a dispatch stall occurs and an oldest instruction in a load queue is a last-level cache miss, a second hardware counter may count total cycles, and a third hardware counter may count committed instructions. Accordingly, a software/firmware mechanism may read the various hardware counters once the committed instruction counter reaches a threshold value and divide a value of the first hardware counter by a value of the second hardware counter to measure a stall fraction during a current program execution phase. The measured stall fraction can then be used to predict a stall fraction in a next program execution phase such that optimal voltage and frequency settings can be applied in the next phase based on the predicted stall fraction.

Abstract: Systems and methods for operating a processor include determining confidence levels, such as high, low, and medium confidence levels, associated with in-flight branch instructions in an instruction pipeline of the processor, based on counters used for predicting directions of the in-flight branch instructions. Numbers of in-flight branch instructions associated with each of confidence levels are determined. A weighted sum of the numbers weighted with weights corresponding to the confidence levels is calculated and the weighted sum is compared with a threshold. A throttling signal may be asserted to indicate that instructions are to be throttled in a pipeline stage of the instruction pipeline based on the comparison.

Abstract: Allocating power between multiple central processing units (CPUs) in a multi-CPU processor based on total current availability and individual CPU quality-of-service (QoS) requirements is disclosed. Current from a power rail is allocated to CPUs by a global current manger (GCM) circuit related to performance criteria set by CPUs. The CPUs can request increased current allocation from the GCM circuit, such as in response to executing a higher performance task. If the increased current allocation request keeps total current on the power rail within its maximum rail current limit, the GCM circuit approves the request to allow the CPU increased current allocation. This can allow CPUs executing higher performance tasks to have a larger current allocation than CPUs executing lower performance tasks without the maximum rail current limit being exceeded, and without having to necessarily lower voltage of the power rail, which could unnecessarily lower performance of all CPUs.

Abstract: The disclosure generally relates to dynamic clock and voltage scaling (DCVS) based on program phase. For example, during each program phase, a first hardware counter may count each cycle where a dispatch stall occurs and an oldest instruction in a load queue is a last-level cache miss, a second hardware counter may count total cycles, and a third hardware counter may count committed instructions. Accordingly, a software/firmware mechanism may read the various hardware counters once the committed instruction counter reaches a threshold value and divide a value of the first hardware counter by a value of the second hardware counter to measure a stall fraction during a current program execution phase. The measured stall fraction can then be used to predict a stall fraction in a next program execution phase such that optimal voltage and frequency settings can be applied in the next phase based on the predicted stall fraction.

Abstract: The disclosure generally relates to dynamic clock and voltage scaling (DCVS) based on program phase. For example, during each program phase, a first hardware counter may count each cycle where a dispatch stall occurs and an oldest instruction in a load queue is a last-level cache miss, a second hardware counter may count total cycles, and a third hardware counter may count committed instructions. Accordingly, a software/firmware mechanism may read the various hardware counters once the committed instruction counter reaches a threshold value and divide a value of the first hardware counter by a value of the second hardware counter to measure a stall fraction during a current program execution phase. The measured stall fraction can then be used to predict a stall fraction in a next program execution phase such that optimal voltage and frequency settings can be applied in the next phase based on the predicted stall fraction.

Abstract: Adaptive voltage modulation circuits for adjusting supply voltage to reduce supply voltage droops and minimize power consumption are provided. In one aspect, an adaptive voltage modulation circuit detects a supply voltage droop by detecting when a supply voltage falls below a droop threshold voltage, and adjusts a clock signal provided to a load circuit in response to a supply voltage droop. The adaptive voltage modulation circuit keeps a count of the number of clock signal cycles during which the supply voltage is below the droop threshold voltage. The adaptive voltage modulation circuit increases the supply voltage in response to the count exceeding an upper threshold value, and decreases the supply voltage in response to the count being less than a lower threshold value at an end of a defined period. The adaptive voltage modulation circuit can reduce the time a load circuit operates with reduced frequency while minimizing power consumption.

Abstract: Systems and methods for managing access to a cache relate to determining one or more execute permissions associated with a write-address of a write-request to the cache. The cache may be a unified cache for storing data as well as instructions. If there is a write-miss in the cache for the write-request, a cache controller may determine whether to implement a write-allocate policy or a write-no-allocate policy for servicing the write-miss, based on the one or more execute permissions. The one or more execute permissions can relate to a privilege level associated with the write-address. Execute permissions of a producing agent which generated the write-request and an execute permission of a consuming agent which can execute from the write-address may be based on the privilege levels of the producing agent and the consuming agent, respectively.

Abstract: Systems and methods for operating a processor include determining confidence levels, such as high, low, and medium confidence levels, associated with in-flight branch instructions in an instruction pipeline of the processor, based on counters used for predicting directions of the in-flight branch instructions. Numbers of in-flight branch instructions associated with each of confidence levels are determined. A weighted sum of the numbers weighted with weights corresponding to the confidence levels is calculated and the weighted sum is compared with a threshold. A throttling signal may be asserted to indicate that instructions are to be throttled in a pipeline stage of the instruction pipeline based on the comparison.

Abstract: Delays due to waiting for operands that will not be used by a select operand instruction, are alleviated based on an early recognition that such operand data is not required in order to complete the processing of the select operand instruction. At appropriate points prior to execution, determinations are made regarding a selection criterion or criteria specified by the select operand instruction, conditions that affect the selection criteria, and the availability of operands. A hold circuit uses the determinations to control the activation and release of a hold signal that controls processor pipeline stalls. A stall required to wait for operand data is skipped or a stall is terminated early, if the selected operand is available even though the other operand, that will not be used, is not available. A stall due to waiting for operands is maintained until the selection criteria is met and the selected operand is fetched and made available.

Abstract: The disclosure generally relates to dynamic clock and voltage scaling (DCVS) based on program phase. For example, during each program phase, a first hardware counter may count each cycle where a dispatch stall occurs and an oldest instruction in a load queue is a last-level cache miss, a second hardware counter may count total cycles, and a third hardware counter may count committed instructions. Accordingly, a software/firmware mechanism may read the various hardware counters once the committed instruction counter reaches a threshold value and divide a value of the first hardware counter by a value of the second hardware counter to measure a stall fraction during a current program execution phase. The measured stall fraction can then be used to predict a stall fraction in a next program execution phase such that optimal voltage and frequency settings can be applied in the next phase based on the predicted stall fraction.

Abstract: The clock frequency of a processor is reduced in response to a dispatch stall due to a cache miss. In an embodiment, the processor clock frequency is reduced for a load instruction that causes a last level cache miss, provided that the load instruction is the oldest load instruction and the number of consecutive processor cycles in which there is a dispatch stall exceeds a threshold, and provided that the total number of processor cycles since the last level cache miss does not exceed some specified number.

Abstract: Circuits, systems, and methods for monitoring a power supply voltage and determining if the power supply voltage has drooped are disclosed. In one embodiment, a voltage monitoring circuit is provided and configured to determine if the power supply voltage supplied to a functional circuit has drooped. When no droop of the power supply voltage is detected, the voltage monitoring circuit is configured to provide an indication to the functional circuit to operate in a first mode. When droop of the power supply voltage is detected, the voltage monitoring circuit is configured to provide an indication to the functional circuit to operate in a second mode. In this manner, operating margin in the power supply may be reduced since the functional circuit may be configured to properly operate when a voltage droop of the power supply voltage occurs.

Abstract: Automatic calibration circuits for operational calibration of critical-path time delays in adaptive clock distribution systems, and related methods and systems, are disclosed. The adaptive clock distribution system includes a tunable-length delay circuit to delay distribution of a clock signal provided to a clocked circuit, to prevent timing margin degradation of the clocked circuit after a voltage droop occurs in a power supply supplying power to the clocked circuit. The adaptive clock distribution system also includes a dynamic variation monitor to reduce frequency of the delayed clock signal provided to the clocked circuit in response to the voltage droop in the power supply, so that the clocked circuit is not clocked beyond its performance limits during a voltage droop. An automatic calibration circuit is provided in the adaptive clock distribution system to calibrate the dynamic variation monitor during operation based on operational conditions and environmental conditions of the clocked circuit.

Abstract: Automatic calibration circuits for operational calibration of critical-path time delays in adaptive clock distribution systems, and related methods and systems, are disclosed. The adaptive clock distribution system includes a tunable-length delay circuit to delay distribution of a clock signal provided to a clocked circuit, to prevent timing margin degradation of the clocked circuit after a voltage droop occurs in a power supply supplying power to the clocked circuit. The adaptive clock distribution system also includes a dynamic variation monitor to reduce frequency of the delayed clock signal provided to the clocked circuit in response to the voltage droop in the power supply, so that the clocked circuit is not clocked beyond its performance limits during a voltage droop. An automatic calibration circuit is provided in the adaptive clock distribution system to calibrate the dynamic variation monitor during operation based on operational conditions and environmental conditions of the clocked circuit.

Abstract: Synthesizing intermediate performance levels in integrated circuits, and related processor systems, methods, and computer-readable media are disclosed. In one embodiment, a synthesized performance level setting circuit receives an input indicating a synthesized performance mode. The circuit generates a power source selection output to select a first power source providing power to an integrated circuit functional block at a first voltage level, and generate a clock frequency setting output to select a first clock frequency associated with the first voltage level to clock the functional block, for a first predefined time interval. The circuit also generates the power source selection output to select a second power source to provide power at a second voltage level lower than the first voltage level, and generate the clock frequency setting output to select a second clock frequency associated with the second voltage level to clock the functional block, for a second predefined time interval.

Abstract: Methods and apparatus for voltage scaling are provided. In an example, an operational limit of a processor is determined by varying a supply voltage to force a processor interrupt fault and/or a processor reset. A clock frequency and the supply voltage can be maintained substantially constant for a time duration. If these operational parameters do not force the processor interrupt fault and/or the processor reset, the supply voltage is varied again, and the clock frequency and the supply voltage are maintained substantially constant for a second time duration. The variation continues until initiation of the processor interrupt fault and/or the processor reset, at which time least one of a clock frequency, the supply voltage, and a temperature are recorded as an operational limit. After determining the operational limit, the supply voltage is adjusted to within the operational limit.

Abstract: Adaptive voltage scalers (AVSs), systems, and related methods are disclosed. The AVSs are configured to adaptively adjust voltage levels powering a functional circuit(s) based on target operating frequencies and delay variation conditions to avoid or reduce voltage margin. In one embodiment, the AVS includes an AVS database. The AVS database can be configured to store voltage levels for various operating frequencies of a functional circuit(s) to avoid or reduce voltage margin. The AVS database allows rapid voltage level decisions. The voltage levels stored in the AVS database may be initial, minimum, learned, populated, explored, backed out, temperature-based, and/or age-based voltage levels according to disclosed embodiments to further avoid or reduce voltage margin. An AVS module may be a software-based module that consults the AVS database to make voltage level decisions. Providing the AVS module as a software-based module may allow flexibility in configuring the AVS module and/or the AVS database.

Abstract: Adaptive voltage scalers (AVSs), systems, and related methods are disclosed. The AVSs are configured to adaptively adjust voltage levels powering a functional circuit(s) based on target operating frequencies and delay variation conditions to avoid or reduce voltage margin. In one embodiment, the AVS includes an AVS database. The AVS database can be configured to store voltage levels for various operating frequencies of a functional circuit(s) to avoid or reduce voltage margin. The AVS database allows rapid voltage level decisions. The voltage levels stored in the AVS database may be initial, minimum, learned, populated, explored, backed out, temperature-based, and/or age-based voltage levels according to disclosed embodiments to further avoid or reduce voltage margin. An AVS module may be a software-based module that consults the AVS database to make voltage level decisions. Providing the AVS module as a software-based module may allow flexibility in configuring the AVS module and/or the AVS database.