Power throttling method and apparatus
View Patent ↗Disclosed is an apparatus which deactivates both the AC as well as the DC component of power for various functions in a CPU. The CPU partitions dataflow registers and arithmetic units such that voltage can be removed from the upper portion of dataflow registers when the software is not utilizing same. Clock signals are also prevented from being applied to these non-utilized components. As an example, if a 64 bit CPU (processor unit) is to be used with both 32 and 64 bit software, the mentioned components may be partitioned in equal sized upper and lower portions. The logic signal for activating the removal of voltage may be obtained from a software-accessible architected control register designated as a machine state register in some CPUs. The same logic may be used in connection with removing voltage and clocks from other specialized functional components such as the floating point unit when software instructions do not presently require same.
1. A method of conserving power in a computer processor, comprising:
reading a software-accessible control register;
determining an idle status of a subunit of the computer processor based on the control register;
providing a clock signal to the subunit based on the determined idle status; and
providing a power voltage to the subunit based on the determined idle status and a power management signal.
2. The method as recited in claim 1 , wherein the control register comprises a plurality of architected bits in a machine state register and at least one bit of the plurality of architected bits is associated with at least one subunit of the computer processor.
3. The method as recited in claim 2 , wherein determining the idle status comprises reading the at least one bit associated with the at least one subunit of the computer processor.
4. The method as recited in claim 1 , further comprising setting one or more of a plurality of bits in the control register based on an idle status of a subunit of the computer processor.
5. The method as recited in claim 1 , wherein the computer processor comprises data flow circuitry comprising a plurality of data flow sections, at least one data flow section configured as a subunit of the computer processor.
6. The method as recited in claim 1 , wherein the computer processor comprises upper and lower bit data register circuitry portions, at least the upper bit data register circuitry portion configured as a subunit of the computer processor.
7. The method as recited in claim 1 , wherein:
the computer processor comprises partitioned dataflow registers comprising a lower portion register consistent in size with the lowest instruction width software to be used in the computer processor; and
the control register comprises an architected control register bit indicating the width of the greatest instruction width presently being used by the computer processor.
8. The method as recited in claim 1 , wherein the computer processor comprises partitioned arithmetic logic units (ALUs), comprising an upper ALU and a lower ALU, at least the upper ALU configured as a subunit of the computer processor.
9. The method as recited in claim 1 , wherein the computer processor comprises a floating point logic unit (FPU), the FPU configured as a subunit of the computer processor.