A compute device may include one or more processors operable at variable performance levels depending upon power supplied from a compute device power supply. A baseboard management controller of the compute device may periodically calculate an adjustment value for the power supply to adjust the power delivered to the one or more processors. The adjustment value may be calculated as a function of a thermal margin between the temperature of the one or more processors over time and a thermal operating limit of the one or more processors.

An electronic device capable of placing restrictions on processor usage is disclosed. The electronic device may include: a memory; and a processor including a first core and a second core. The memory may store instructions that, when executed by the processor, cause the first core to transition from an active state to an idle state in response to a restriction signal for the first core, and cause the first core to transition to a power save state when the first core remains in the idle state for at least a preset time. For hot-unplugging, as the electronic device does not transition a core to an offline state, it does not have to perform cleanup operation on the memory and variables. Hence, it is possible to reduce the latency time due to hot-unplugging.

According to at least one example embodiment, a method and corresponding apparatus for controlling power in a multi-core processor chip include: accumulating, at a controller within the multi-core processor chip, one or more power estimates associated with multiple core processors within the multi-core processor chip. A global power threshold is determined based on a cumulative power estimate, the cumulative power estimate being determined based at least in part on the one or more power estimates accumulated. The controller causes power consumption at each of the core processors to be controlled based on the determined global power threshold. The controller may directly control power consumption at the core processors or may command the core processors to do so.

Disclosed is an electronic device which includes a main processor, and a systolic array processor, and the systolic array processor includes processing elements, a kernel data memory that provides a kernel data set to the processing elements, a data memory that provides an input data set to the processing elements, and a controller that provides commands to the processing elements. The main processor translates source codes associated with the systolic array processor into commands of the systolic array processor, calculates a switching activity value based on the commands, and stores the translated commands and the switching activity value to a machine learning module, which is based on the systolic array processor.

Systems, apparatuses, and methods for managing power consumption for a neural network implemented on multiple graphics processing units (GPUs) are disclosed. A computing system includes a plurality of GPUs implementing a neural network. In one implementation, the plurality of GPUs draw power from a common power supply. To prevent the power consumption of the system from exceeding a power limit for long durations, the GPUs coordinate the scheduling of tasks of the neural network. At least one or more first GPUs schedule their computation tasks so as not to overlap with the computation tasks of one or more second GPUs. In this way, the system spends less time consuming power in excess of a power limit, allowing the neural network to be implemented in a more power efficient manner.

A system for providing system level sleep state power savings includes a plurality of memory channels and corresponding plurality of memories coupled to respective memory channels. The system includes one or more processors operative to receive information indicating that a system level sleep state is to be entered and in response to receiving the system level sleep indication, moves data stored in at least a first of the plurality of memories to at least a second of the plurality of memories. In some implementations, in response to moving the data to the second memory, the processor causes power management logic to shut off power to: at least the first memory, to a corresponding first physical layer device operatively coupled to the first memory and to a first memory controller operatively coupled to the first memory and place the second memory in a self-refresh mode of operation.

A computing device includes an array of processing elements mutually connected to perform single instruction multiple data (SIMD) operations, memory cells connected to each processing element to store data related to the SIMD operations, and a cache connected to each processing element to cache data related to the SIMD operations. Caches of adjacent processing elements are connected. The same or another computing device includes rows of mutually connected processing elements to share data. The computing device further includes a row arithmetic logic unit (ALU) at each row of processing elements. The row ALU of a respective row is configured to perform an operation with processing elements of the respective row.

In one embodiment, a processor includes a plurality of cores each including a first storage to store a physical identifier for the core and a second storage to store a logical identifier associated with the core; a plurality of thermal sensors to measure a temperature at a corresponding location of the processor; and a power controller including a dynamic core identifier logic to dynamically remap a first logical identifier associated with a first core to associate the first logical identifier with a second core, based at least in part on a temperature associated with the first core, the dynamic remapping to cause a first thread to be migrated from the first core to the second core transparently to an operating system. Other embodiments are described and claimed.

Embodiments include an apparatus comprising an execution unit coupled to a memory, a microcode controller, and a hardware controller. The microcode controller is to identify a global power and performance hint in an instruction stream that includes first and second instruction phases to be executed in parallel, identify a local hint based on synchronization dependence in the first instruction phase, and use the first local hint to balance power consumption between the execution unit and the memory during parallel executions of the first and second instruction phases. The hardware controller is to use the global hint to determine an appropriate voltage level of a compute voltage and a frequency of a compute clock signal for the execution unit during the parallel executions of the first and second instruction phases. The first local hint includes a processing rate for the first instruction phase or an indication of the processing rate.

A memory chip includes at least two memory blocks. In a method for controlling power supply for the memory blocks of the memory chip, each memory block receives a command for switching to standby mode. The commands are issued, for example by a processor, separately for each memory block in order to be able to individually place the memory block in standby mode.