Patent Application
20170075397
This application relates to thermal mitigation and, more specifically, to providing thermal mitigation to a computing device in response to measured power consumption of the device.
A conventional modern smart phone may include a system on chip (SOC), which has a processor and other operational circuits. Specifically, an SOC in a smart phone may include a processor chip within a package, where the package is mounted on a printed circuit board (PCB) internally to the phone. The phone includes an external housing and a display, such as a liquid crystal display (LCD). A human user when using the phone physically touches the external housing and the display.
As the SOC operates, it generates heat. In one example, the SOC within a smart phone may reach temperatures of 80° C.-100° C. Furthermore, conventional smart phones do not include fans to dissipate heat. During use, such as when a human user is watching a video on a smart phone, the SOC generates heat, and the heat is spread through the internal portions of the phone to the outside surface of the phone.
The outside surface of the phone is sometimes referred to as the “skin.” The outside surface includes the part of the external housing that is physically on the outside of the phone as well as any other externally-exposed portions, such as an LCD display. It is generally accepted that the skin of the phone should not reach temperatures higher than about 40° C.-45° C. due to safety and ergonomic reasons. As noted above, the SOC within the smart phone may reach temperatures of 80° C.-100° C., although the temperature of the SOC is not felt directly at the skin of the phone. Instead, heat dissipation within the phone often means that the skin temperature of the phone is at a lower temperature than the SOC temperature. Furthermore, whereas changes to SOC temperature may be relatively quick (e.g., seconds), changes to device skin temperature may be relatively slow (e.g., tens of seconds or minutes).
Conventional smart phones include algorithms to control the skin temperature by reducing a frequency of operation of the SOC when a temperature sensor on the SOC reaches a threshold level. However, SOC temperature can be a poor proxy for device skin temperature.
Various embodiments include systems and methods that mitigate temperature by measuring power consumption and reducing performance of a processor, if appropriate, based at least in part on the power consumption measurement.
In one embodiment, a method includes generating a power consumption reading that is indicative of power consumption of a computing device. For instance, the computing device may be a smart phone, tablet, or other device. The example method also includes comparing the power consumption reading to a power threshold, wherein the power threshold represents a level of power consumption corresponding to a rise in temperature of an external surface of the computing device. In response to determining that the power consumption reading exceeds the power threshold, the method further includes measuring cumulative power consumption over time from the power consumption reading. The cumulative power consumption over time is compared to an energy threshold, and in response to determining that the cumulative power consumption over time exceeds the energy threshold, the method includes reducing an operating parameter of the device to reduce power consumption. For instance, the operating parameter may include frequency of operation, voltage of operation, or other parameters.
In another embodiment, a system includes a computer processor configured to execute machine-readable instructions and to consume power from a system battery, and a physical housing that encloses at least part of the system. The computer processor is enclosed within the physical housing and is in thermal contact with the physical housing. The system also includes a power indicating device disposed within the physical housing, where examples of power indicating devices include current measurement devices and power measurement devices. The processor is configured to calculate a cumulative power consumption from a reading of the power indicating device, compare the cumulative power consumption over time of the system to an energy threshold, and reduce an operating parameter of the computer processor in response to determining that the cumulative power consumption over time of the system exceeds the energy threshold.
In yet another embodiment, a computer program product having a computer readable medium tangibly recording computer program logic for mitigating temperature of a computing device includes code to compare a power consumption of the computing device to a power threshold corresponding to a rise in temperature of an external surface of the computing device. The computer program product also includes code to compare a cumulative power consumption of the computing device to an energy threshold, corresponding to a temperature threshold of the external surface of the computing device, in response to determining that the power consumption exceeds the power threshold, and code to reduce a frequency of operation of the computing device in response to determining that the cumulative power consumption exceeds the energy threshold.
Various embodiments provided herein include systems and methods to mitigate temperature of a computing device. For instance, various embodiments described herein use power measurement in addition to, or instead of, temperature measurement as a basis for a thermal mitigation algorithm.
In one embodiment, a computer processor includes a system on chip (SOC) that has a number of processing cores and is associated with a power supply. The SOC and the power supply are disposed within a computing device, such as a smart phone. The smart phone also includes a battery, a printed circuit board hosting the SOC and power supply, a touchscreen display, and an outer housing, among other things. The power supply converts the voltage and current from the battery into a voltage and current that can be used by the SOC. As the SOC operates, it produces heat.
The heat from the SOC spreads throughout the phone according to the heat conduction properties of the physical materials that make up the phone. However, it is generally expected that the skin of the phone will not be as hot as the SOC, at least under usual operating conditions. The system includes one or more processes that monitor power use of the computer processor. For instance, in one embodiment a current sensor on a battery rail may measure the current, and power is determined from the current since the voltage is known. In another embodiment, one or more current sensors between the power supply and the SOC may measure current consumption by the SOC, which can be translated into power when voltage is known. In yet another embodiment, a current sensor specific to one or more cores of the SOC may provide an indication of current consumption by those one or more cores. Once again, current consumption can be translated into a power reading when voltage is known. In fact, the scope of embodiments may include measuring current consumption at any appropriate component of the system. In another embodiment, one or more digital power meters (DPMs) may be used at any appropriate part of the system to provide an indication of power consumption.
A process, either hardware-based or software-based or combination thereof, uses the power reading for thermal mitigation. Specifically, in this example, the process compares the power reading to a first threshold to determine whether the power reading goes above the first threshold. If the power reading stays below the first threshold, then the process keeps monitoring without providing thermal mitigation action. If the power reading is above the first threshold, then the process moves to another action that includes measuring a cumulative power consumption over time. If the cumulative power consumption over time crosses a second threshold, then the process begins thermal mitigation by, e.g., reducing a clock frequency of the processor. If the cumulative power consumption over time does not cross the second threshold, then the process continues to monitor the cumulative power consumption over time until the measured power consumption drops below the first threshold.
The process described above may be embodied as computer executable code that is read from a tangible computer-readable medium and executed by a kernel process of the processor. In another embodiment, the process may be embodied as a hardware process built in to the processor. However, in many embodiments, thermal changes at the skin of the device and at the SOC change on the order of seconds or minutes, so that software is generally fast-acting enough.
Other embodiments include methods to perform the process described above. For example, a method may include generating a power consumption reading in the device and comparing it to a first power threshold. In response to the power reading exceeding the first power threshold, the processor compares the cumulative power consumption over time to a second threshold. In response to the cumulative power consumption over time exceeding the second threshold, the processor reduces its operating frequency, thereby reducing the amount of power consumed and heat generated.
As shown in
As a computer processor operates, it produces heat, which dissipates throughout the physical structure of computing device 100. Depending on the specific thermal properties of computing device 100, heat from the operation of the processor may reach uncomfortable or near-uncomfortable temperatures on the outside surface 120 of computing device 100. The computer processor within computing device 100 provides functionality to control the heat felt on the outside surface of the computing device 100 by measuring power consumption and adjusting a frequency and/or voltage of the processor if appropriate.
Battery 205, power management integrated circuit (PMIC)210, and SOC 230 are disposed within the computing device 100 so that they are enclosed within the physical housing of the computing device 100 as indicated by exterior surface 120. Battery 205, PMIC 210, and SOC 230 are also in thermal contact with the physical housing so that the heat generated by those items is conducted to the exterior surface 120 of the device 100.
Computing device 100 includes battery 205, which may be any appropriate battery now known or later developed. For instance, battery 205 may include a lithium-ion battery or other power source. Battery 205 is coupled to battery rail 206, which distributes power from the battery 205. Battery rail 206 is coupled to device display 110 and to power management integrated circuit (PMIC) 210. In some embodiments, the power from battery rail 206 may be regulated or otherwise conditioned before being provided to display 110, however for ease of illustration, battery rail 206 in this example is provided directly to display 110. Current measurement device 220a is provided to measure the current on battery rail 206.
PMIC 210 receives power from battery rail 206 and regulates the voltage to provide an output voltage that is usable by SOC 230. In this example, current measurement device 220b is placed on the battery rail in a location so that it measures an amount of current provided to PMIC 210 but does not directly measure current provided to display device 110. Therefore, current measurement device 220b may be used to provide an indication as to power consumption of SOC 230.
SOC 230 has four cores, 231-234. Examples of cores include a central processing unit (CPU), a graphics processing unit (GPU), a modem, and the like. Although not shown in
Further shown in
Additionally, this embodiment also includes current measurement devices 220d-g, where each one of the devices 220d-g is associated with a particular one of the cores 231-234. Each of the current measurement devices 220d-g may be included for instance in a global distributed head switch (GDHS) associated with its corresponding core 231-234. For instance, some conventional systems may include a GDHS with each of the cores 231-234, and various embodiments described herein may use the current measurement functionality already included in the GDHSs.
SOC 230 includes functionality to derive power use from various current measurement data provided by current measurement devices 220. The architecture of
The scope of embodiments is not limited to current measurements only. Rather, other embodiments may include one or more DPMs placed at any of a variety of locations to measure power consumption on the battery rail as a whole or to various functional units (e.g. display 110 and SOC 230) individually. In an embodiment wherein current measurements are provided to the thermal mitigation algorithm, the thermal mitigation algorithm may derive power readings by for instance multiplying current by voltage. On the other hand, DPMs may provide an indication of power to the thermal mitigation algorithm.
SOC 230 includes a software kernel 330, which operates when SOC 230 is powered up and receiving a clock signal. Thermal mitigation unit 335 is a software process that is built into kernel 330, and it performs the method described with respect to
Thermal mitigation unit 335 receives one or more current measurements from current measurement devices. Examples of current measurement devices include those shown in
Thermal mitigation unit 335 may reduce the clock frequencies provided to the cores or increase the clock frequencies provided to the cores by sending commands to clock control unit 312. Clock control unit 312 may be physically a part of SOC 230 or separate therefrom, as the scope of embodiments is not limited to any particular clocking architecture. Clock control unit 312 may control for instance a phase locked loop (PLL) or other appropriate circuit that provides a periodic clock signal in order to raise or lower the operating frequency of one or more of the cores 231-234.
On the other hand, some computing devices may have different levels of heat spreading, such that heat from the SOC is conducted from the SOC less evenly. Therefore, the heat is concentrated to specific areas of the skin of the device. Such areas may become heated more quickly, and the heat may be removed from the surface by ambient air less efficiently since less surface area is heated. Such a computing device may utilize thermal mitigation algorithms more often in order to avoid heating those specific areas beyond a level that is comfortable for a human user.
Each model of computing device has its own thermal properties. Therefore, a thermal algorithm that is designed specifically for a particular model of computing device may not work well for different models of computing devices. Returning to the graph of
In some embodiments, Pth_skin may be a design parameter that is known by simulation during the design phase of the device. In other embodiments, Pth_skin may be determined through experimentation with a physical embodiment of the device. For instance, a computer-aided design program may be used by a designer to determine the heat conductive properties of the device as the devices being designed. Additionally or alternatively, a designer may test power consumption and take temperature readings of the skin of the device in a controlled environment in order to determine a value for Pth_skin. Such design and/or testing may be performed by a designer or manufacturer of the device.
Once power consumption exceeds the threshold Pth_skin, power use above that threshold is expected to increase the skin temperature up to an uncomfortable level, unless power use is decreased. Higher levels of power use are also expected to increase skin temperature more quickly than lower levels of power use. Accordingly, curve 410 represents a boundary above which thermal mitigation is invoked according to the method of
As with Pth_skin, the level of the boundary of curve 410 is specific to a particular device model and type. In this example, the level of the boundary of curve 410 corresponds to a skin temperature threshold so that a cumulative power consumption over a certain amount of time, exceeding the boundary of the curve 410, would be expected to cause the skin temperature of the device to reach a pre-defined uncomfortable level, such as 40° C.-45° C. The level of the boundary of curve 410 depends upon the specific thermal conductive properties of a particular device. Those thermal conductive properties may be known through design and/or experimentation. Therefore, a designer using computer-aided design may determine the boundary of curve 410, or the designer may use experimentation on a physical implementation of the device to determine the boundary of curve 410. While the boundary of curve 410 is shown as a curve approaching zero as power increases, the scope of embodiments may include any appropriately shaped curve that uses power consumption as a proxy for skin temperature of the device. For example, in other embodiments, the relationship expressed by curve 410 may be linear.
In one example use case, a manufacturer or designer of a computing device determines Pth_skin and the boundary of curve 410. The designer or manufacturer then saves that information into memory of SOC 230 for use by thermal mitigation unit 335. This process is performed before delivering finished units to consumers, so that the finished product includes robust thermal mitigation functionality built into it.
A flow diagram of an example method 500 of providing thermal mitigation is illustrated in
At action 501, the thermal mitigation unit 335 reads temperature sensing data (TSENS) and a power sensing data. As noted above, power sensing data may be derived from one or more current measurements or may be provided directly by a power measurement sensor. The rest of this description will focus on power measurement and will not consider the temperature sensor reading itself. Nevertheless, it is understood that various embodiments may be performed in addition to other processes that use temperature sensing.
At action 502, the thermal mitigation unit 335 compares the measured power reading (e.g., in Watts) against a power threshold, Pth_skin. The value of the power threshold may be any appropriate amount, and it represents a level of power consumption that may be expected to begin an undesirable rise in device skin temperature. The value of the power threshold will depend on the heat conduction properties of the particular device. For instance, a device with a heat spreading layer built into it may be assigned a higher power threshold than a device that allows a more rapid heat transfer from the SOC to the skin. In some scenarios, the power threshold may be assigned to a device based upon experimentation and/or known heat transfer properties of the design. As noted above, Pth_skin may be saved to memory in the processor of the device and accessed by thermal mitigation unit 335 as it performs method 500.
If the measured power reading is not greater than the power threshold, then the thermal mitigation unit 335 continues to monitor the power sensor. However, if the measured power reading is greater than the power threshold, the thermal mitigation unit 335 moves to action 503. At action 503 the thermal mitigation process measures power consumption over time. An example includes integrating power consumption over a particular time period. Another example is a cumulative, moving window average of power consumption over time.
At action 504, the thermal mitigation process determines whether the power consumption over the time period is greater than an energy threshold (Ethreshold), where the energy threshold corresponds to amount of energy expected to cause an undesirably high device skin temperature. (e.g., 43° C.). The energy threshold is specific to a particular device design, and may be assigned to a particular device through experimentation and/or known heat transfer properties of the design. In the particular example of
In one particular use case, the action 504 performs the following calculation according to equation 1:
E/ΔT>E
threshold
/ΔT
device, (1)
where ΔT is the amount of time over which the calculation of action 503 is integrated, and wherein ΔTdevice is determined by the shape of the curve 410. For instance, the calculation of action 503 may be performed and accumulated at predefined times, so that the value of E gets larger as time continues and as long as the power measurement does not drop below Pth_skin. The calculation of action 504 divides E by that window of time that has elapsed and compares it to Ethreshold normalized by ΔTdevice. Therefore, the actions of blocks 503 and 504 calculate a cumulative power consumption over time and compare the cumulative power consumption to an energy threshold normalized by ΔTdevice to provide a threshold power consumption over time. In another embodiment, action 503 includes calculating a cumulative, moving window average of power consumption over a defined time period.
If the power consumption over the time period is less than the energy threshold, then the thermal mitigation unit 335 continues to monitor the power consumption. If the power consumption over time is greater than the energy threshold, then the thermal mitigation unit 335 moves to action 505.
At action 505, the thermal mitigation unit 335 reduces the energy consumed by the SOC. In one example, the thermal mitigation process reduces an operating frequency of one or more cores in the SOC, thereby reducing power consumption. However, the scope of embodiments may include any appropriate thermal mitigation technique, such as putting cores in an idle state. For instance, in the example of
The scope of embodiments is not limited to the specific method shown in
Furthermore, some embodiments may use power measurements for any relevant portion of the computing device. For example, as shown above in
Various embodiments may provide one or more advantages over conventional solutions. For instance, it may be difficult to capture a temperature reading directly from the skin of a computing device, especially for more compact and mobile computing devices such as phones and tablets. Nevertheless, skin temperature can be very relevant to a user's perception of comfort. Some conventional solutions use temperature readings gathered from sensors on the processor and base thermal mitigation decisions on that temperature reading. But temperature readings gathered from thermal sensors on the processor may not provide an accurate indication of skin temperature, thereby causing intervention of a thermal mitigation process too early or too often and sacrificing performance of the system.
By contrast, the systems described herein provide thermal mitigation using power consumption as a proxy for temperature. Ultimately, temperatures affecting the skin of the device are generally expected to be caused by power consuming units within the computing device, and the embodiments described above provide a solution to monitor power consumption and to provide thermal mitigation in view of the power consumption of the device. When paired with device-specific data to provide the power threshold and a mitigation curve, such as that shown in
As those of some skill in this art will by now appreciate and depending on the particular application at hand, many modifications, substitutions and variations can be made in and to the materials, apparatus, configurations and methods of use of the devices of the present disclosure without departing from the spirit and scope thereof. In light of this, the scope of the present disclosure should not be limited to that of the particular embodiments illustrated and described herein, as they are merely by way of some examples thereof, but rather, should be fully commensurate with that of the claims appended hereafter and their functional equivalents.
