Miscellaneous active electrical nonlinear devices – circuits – and – Specific identifiable device – circuit – or system – With specific source of supply or bias voltage
Reexamination Certificate
2001-03-16
2003-01-21
Zweizig, Jeffrey (Department: 2816)
Miscellaneous active electrical nonlinear devices, circuits, and
Specific identifiable device, circuit, or system
With specific source of supply or bias voltage
Reexamination Certificate
active
06509788
ABSTRACT:
TECHNICAL FIELD
This invention relates in general to management of power consumption by an integrated circuit, and in specific to a system and method that utilize an on-chip voltage controlled oscillator to dynamically generate an appropriate clock frequency for the chip based on the voltage being drawn by the chip.
BACKGROUND
Integrated circuits (commonly referred to as “chips”), such as microprocessors, are utilized in an ever-increasing number of various applications. For instance, such chips are commonly implemented not only in personal computers (PCs) and laptops, but are typically implemented in much smaller (and more portable) devices, such as personal digital assistants (PDAs), cellular telephones, pagers, and various other types of devices. Considering the number of tasks that such chips are relied upon to perform, the desire for fast processing speeds (to allow tasks to be performed quickly), and the desire for limited power consumption by such chips, chip designers are faced with the difficult task of designing chips that achieve the desired performance (e.g., fast processing speed), while managing the power consumption of the chips. Given the ever-increasing advances being made in performance of chips, such as microprocessors, power consumption is becoming a serious concern. For example, power consumption is becoming a serious performance limiter for high speed microprocessors. For instance, a key design objective for microprocessor systems is providing the highest possible peak performance for compute-intensive code, while reducing power consumption of the microprocessor system. Particularly when such microprocessor systems are to be implemented within portable electronic devices, reduction in power consumption (at least during low performance periods) is desirable to maximize the battery life of the device.
As is well known, power consumption of a chip may be generally computed utilizing the following equation: P=C*V
2
*F, wherein P represents power consumption, C represents switching capacitance, V represents operating voltage, and F represents the clock frequency of the chip. In view of such equation, it should be understood that switching capacitance (C), voltage (V), and frequency (F) are all factors in determining the power consumption (P) of a chip. In many cases, it is necessary to limit processor frequency (F) in order to hold the power consumption (P) of a chip below a certain level that is acceptable for use in a given system (e.g., within a desktop or portable devices).
Microprocessor chips of the prior art have typically been implemented with a fixed voltage and frequency determined to prevent the chip from consuming more than a particular amount of power. Typically, in designing prior art microprocessor chips, a designer tests the chip with software code for creating a heavy computational load on the chip in order to determine the appropriate voltage and frequency that may be implemented for the chip such that its power consumption does not exceed a particular amount when heavy computational loads are encountered by the chip. However, once implemented, such heavy computational loads may be encountered relatively seldom, with low (or no) computational load being placed on the microprocessor much of the time. Accordingly, the worst case computational loads dictate the voltage and frequency of the chip, thereby hindering performance of the chip (e.g., because of the decreased frequency required for the worst case).
Another common power-saving technique of the prior art attempts to reduce only the clock frequency (F) during non-compute intensive activity. This reduces power, but does not affect the total energy consumed per process. That is, a reduction in frequency (F) results in a linear reduction in the power consumed, but also results in a linear increase in task run-time, which causes the energy-per-task to remain constant. On the other hand, reducing only the voltage (V) of the processor improves its energy efficiency, but compromises its peak performance. It has been recognized in the prior art that if clock frequency (F) and supply voltage (V) are dynamically varied in response to computational load demands, then energy consumed per process can be reduced for the low computational periods, while retaining peak performance when required (i.e., for heavy computational periods). Design strategies attempting to utilize such dynamic variation of clock frequency (F) and supply voltage (V) based on computational loads are commonly referred to as dynamic voltage scaling (DVS).
One implementation proposed in the prior art utilizes DVS on a microprocessor under direct Operating System (OS) control. In such an implementation, one or more voltage scheduler algorithms are required in the OS of a DVS system, which are utilized to dynamically adjust the processor speed and voltage at run-time of a microprocessor. The voltage schedulers control the clock frequency (F) and supply voltage (V) of a microprocessor by writing a desired frequency (in MHz) to a coprocessor register. The voltage schedulers analyze the current and past state of the system in order to predict the future workload of the processor. For example, individual applications supply a completion deadline, and the voltage scheduler uses the applications' previous execution history to determine the number of processor cycles required and sets the clock frequency (F) accordingly. Interval-based voltage schedulers have been proposed in the prior art, which periodically analyze system utilization to control the frequency and voltage. As an example, if the voltage scheduler determines that the preceding time interval was greater than 50% active, it may increase the frequency and voltage for the next time interval. Thus, the system attempts to preserve the amount of power consumed by a microprocessor by having the OS dynamically adjust the clock frequency (F) to the minimum level required by the current active processes. Such an implementation that utilizes voltage scheduler algorithms (i.e., software) in the OS to dynamically control the voltage and frequency of a microprocessor is described in greater detail in
A Dynamic Voltage Scaled Microprocessor System
, by Thomas Burd, Trevor Pering, Anthony Sratakos, and Robert Brodersen, published in
Journal of Solid State Circuits
Vol. 35, No. 11, November 2000, and
The Technology Behind Crusoe™ Processors
, by Alexander Klaiber (Transmeta Corporation), available (as of the filing of this application) at http://www.transmeta.com/crusoe/download/pdf/crusoetechwp.pdf, the disclosures of which are hereby incorporated herein by reference.
However, such an approach that utilizes the OS to dynamically control the voltage and frequency of a microprocessor is often problematic/undesirable. First, changing a system's OS to implement such an approach is typically very time consuming and/or costly. System administrators generally dislike upgrading their OS to improve their hardware. Additionally, the OS approach is not perfectly reliable because it has imperfect information about how much power the chip is actually consuming and what its compute needs are. Rather, the OS can only attempt to estimate/guess what is needed at the chip level. Furthermore, data necessary for the OS to intelligently estimate the power consumption and/or compute needs of a chip is typically chip specific, which results in greater difficulty in implementing/upgrading such an OS approach (because the OS implementation must be tailored to a specific chip technology that is implemented).
Yet another power-saving technique implemented in prior art microprocessor chips involves regulating the voltage of the on-chip power supply to reduce voltage droops, thereby allowing for higher clock frequency. More specifically, this technique attempts to improve the integrity of the on-chip power supply by regulating it carefully so that the average voltage to the chip can be reduced (i.e., because sudden changes in power consumption tend to cause the voltage to droop below average). By redu
Josephson Don D
Naffziger Samuel D.
Hewlett--Packard Company
Zweizig Jeffrey
LandOfFree
System and method utilizing on-chip voltage controlled... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method utilizing on-chip voltage controlled..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method utilizing on-chip voltage controlled... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3069350