US 5,355,501 AGrant
Idle Detection System
Issue Date:1994-10-11
•65 Claims
•5 Drawing Sheets
Abstract
A method and apparatus for idle detection in a computer system. The operating system of the computer includes an idle kernel which performs idle detection to detect idle states when application program processes are idle. To check for idle states, the idle kernel monitors operating system functions which are typically used by application programs when they are otherwise idle. The operating system includes an idle driver which controls idle operations in response to idle states detected by the idle monitor. The idle driver causes the system to enter the power saving mode with frequent, sometimes short idle intervals.
Metadata
Assignee
- Novell, Inc.
Inventors
- Roger A. Gross
- John P. Constant
Application Information
Application Number:US 491211&
Filing Date:1990-03-09
Priority Date:1990-03-09
Art Unit:237
Classifications
IPC:
G06F 1500G06F 1130
Field of Search:
395364371750;700;57570716.3
Patent Drawings (5 sheets)
Description
BACKGROUND OF THE INVENTION
A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
1. Field of the Invention
The present invention relates to computers and particularly to methods and apparatus for idle detection and power saving in computers.
2. Description of the Related Art
The need for idle-detection arises when, for example, a computer or part thereof is to be slowed down or stopped in order to conserve power. The need to save power is particularly important in battery-operated computers in order to extend battery life. Portable computers are typically battery-operated and hence the need to save power is becoming increasingly important as the use of portable computers increases.
Power consumption is distributed among the major parts of a computer. One part of the computer with significant power consumption is the CPU. Other parts include each of the input/output (I/O) devices such as display screens, keyboards, modems, printers, disk drives and the like. Still another part with significant power consumption is memory.
Prior art attempts at conserving power have generally focused on a single part of the computer such as a particular one of the I/O devices.
For example, to save power for the screen display, systems have detected screen inactivity and employed screen blanking for an idle screen to reduce the power to the display screen when the screen has not been used for some period of time. Typically, a timeout circuit senses changes in screen information and, if no change has occurred for a predetermined timeout period, the backlight to the screen is turned off for power reduction. While screen blanking is effective in reducing power for the display screen, no reduction results in power to the driver circuitry for the display, to the CPU, or to other parts of the computer.
Other prior art attempts at conserving power consumption have focused on disk drives because the power consumption of rotating magnetic disks is high. Disk drive manufacturers have employed various schemes for reducing the power consumption of the disk drive. While such power consumption schemes are effective for the disk drive, no reduction results in power to the CPU or other parts of the computer.
Other attempts at conserving power have done so only when relatively long periods of idleness exist.
In order to reduce power consumption in computers, there is a need for improved idle detection methods and apparatus.
SUMMARY OF THE INVENTION
The present invention is a method and apparatus for idle detection in a computer system. The computer system typically includes as hardware a central processing unit (CPU), memory (M) and input/output (I/O) devices. The computer system additionally includes as software an operating system, DOS, and application programs (AP) that execute in the computer system under control of the operating system.
The operating system includes an idle kernel which performs idle detection to detect idle states when application program processes are idle. A process is idle while it is waiting for some event to occur, for example, an external event such as a keystroke, a mouse movement or a fixed amount of time to pass. To check for idle states, the idle kernel monitors operating system functions which are typically used by application programs when they are otherwise idle.
The operating system includes an idle driver which controls idle operations in response to idle states detected by the idle monitor. The idle driver can be stored in read-only memory (ROM) or otherwise in the system memory. The idle driver causes the system to enter the power saving mode with frequent, sometimes short idle intervals. The idle intervals, for example, can be as short as the time between events, such as between two keystrokes on a keyboard. Even though the idle intervals are at times short, the accumulated power savings of many short idle intervals is substantial. Substantial power saving occurs even when no long idle intervals exist. For example, a long idle interval is measured in multiple seconds or minutes.
The idle kernel and the idle driver use a common data region in memory to pass information about the current application process state and modify the action of the idle detection code.
The idle driver qualifies the idle status information received from the idle kernel. This qualification is done, for example, by checking for specific interrupts. The idle driver discriminates between an application program that is genuinely idle, for example when polling the keyboard in a tight loop, and an application program that is active but also polling the keyboard, for example when testing for an abort key to be pressed. This discrimination is achieved by monitoring the time taken to go idle. If the time is within a specified period, it is assumed that the program is idle polling in a tight loop for a key to be pressed. If the time is outside of the specified limit, then it is assumed that some processing has occurred in between polling the keyboard, and the idle driver does not go to the idle state but returns immediately to the active state.
A program is set to the idle condition when an idle state has been detected within the time period specified in a local idle variable.
In one embodiment, the local idle variable is dynamically calculated as a function of the performance of the computer system. If the local idle variable is IDLE.sub.-- CNTDN, in one example, the initial value is dynamically calculated as follows:
where,
INT16.sub.-- TIME is the time taken to perform a ROS INT16h function 1 (keyboard status check).
RTC.sub.-- TIME is the time taken to perform a read of the real-time clock using the ROS INT1 Ah function 2 (read real time clock).
While measurement of the combination of a real-time clock and interrupt times is one way of measuring performance, other measurements to set the idle variable IDLE.sub.-- CNTDN are, of course, possible.
Since performance varies from computer to computer, a dynamically calculated value for IDLE.sub.-- CNTDN results in improved performance of the idle detection system. The value is calculated during execution of a start-up sequence to determine an initial value and periodically during the execution of application programs to obtain updated values. Updated values are useful since, for example, application programs may filter INT16 and RTC calls and change their duration. The times are calculated, in the example given, by reading the timer I/O device at the start and end of the ROS calls, and calculating the elapsed time.
In accordance with the above summary, the present invention achieves the objective of providing an improved idle detection system.
The foregoing and other objects, features and advantages of the invention will be apparent from the following detailed description in conjunction with the drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 depicts a block-diagram representation of a computer system with idle detection in accordance with the present invention.
FIG. 2 depicts a block-diagram representation of the software for the computer of FIG. 1.
FIG. 3 depicts a block-diagram representation of one embodiment of the memory space of the computer system of FIG. 1 and FIG. 2.
FIG. 4 depicts a block-diagram representation of another embodiment of the memory space of the computer system of FIG. 1 and FIG. 2.
FIG. 5 depicts a block-diagram representation of the idle detection system software of the computer system of FIG. 1 and FIG. 2.
FIG. 6 depicts a block-diagram representation of a text editor application program which executes with the idle detection system of the present invention.
FIG. 7 depicts a flow-chart representation of the operation of the computer system during the execution of the text editor application program of FIG. 6.
DETAILED DESCRIPTION
In FIG. 1, a computer system incorporating the idle detection system of the present invention is shown. The idle detection system is distributed throughout the hardware of the FIG. 1 computer and is stored in the software for the FIG. 1 computer as shown in FIG. 2. The software includes the idle kernel 90 and the idle driver 79, stored for example, all or in part in memory 13 and memory 14.
In FIG. 1, computer 5 includes a central processing unit (CPU) 10, an I/O bus 11, a memory 12 including read/write (R/W) memory 13 and read-only memory 14, and a plurality of I/O devices 30 and controllers therefor 20. The I/O devices 30 include a floppy disk 31 and controller 21, a hard disk 22 and controller 32, a monitor 33 and controller 23, a keyboard 34 and controller 24, a communication device (COM) 36 and controller 25, a display 36 and controller 26, a real time clock (RTC) 37 and controller 27, system timer 38 and controller 28, and a network device (NET) 39 and controller 29. While many of the I/O devices of FIG. 1 may not be present in any particular computer system which employs the present invention, such devices a common along with others not shown.
In FIG. 1, computer 5 also includes a clock unit 6 which provides the CLK signal to the CPU 10. The clock unit 6 is an I/O device which controls the ON and OFF state of and the frequency of the CLK signal on line 7 to the CPU 10. The clock signal is typically employed to control the refresh timing of dynamic memory and other components in a conventional computer 5.
In FIG. 2, a block-diagram representation of the software for the computer of FIG. 1 is shown. In FIG. 2, the software includes the operating system which typically includes as components COMMAND.COM 60, BDOS 61, BIOS 62 and ROM BIOS 63. In an IBM compatible computer, the components 60, 61, and 62 are the functional equivalent of the well-known COMMAND.COM, IBMDOS.COM, and the IBMBIOS.COM. The operating system thus configured, interacts with an application program 59 which together execute on the computer 5 of FIG. 1. In the present invention, the BDOS 61 includes an idle kernel 90 and the BIOS 62 includes an idle driver 79 which together with the rest of the system perform the idle detection of the present invention.
The general operation of the FIG. 1 and FIG. 2 system is well-known. Whenever computer 5 is rebooted, such as upon a power-off to power-on transition, path 72 sends a reboot signal to the ROM BIOS 63 which in turn sends a boot signal to BIOS 62, which in turn sends a boot signal to COMMAND.COM which then boots the system. Both COMMAND.COM 60 and application 59 execute in a conventional manner and call BDOS in the operating system by issuing an interrupt (specifically INT21). BDOS 61 in turn calls routines in the BIOS 62 which may in turn call the ROM BIOS 63 whenever an I/O operation is to occur. Calls to the ROM BIOS typically use interrupts INT10 and INT1F for IBM compatible computers. The BIOS 62 and/or the ROM BIOS 63 in turn control communication between the I/O controllers 20 and devices 30 of the computer 5 of FIG. 1.
In FIG. 3, the location in memory of the various software components is represented. Typically, the interrupt vectors are in the lowest address space. Next, in one embodiment, in order from lower to higher is the R/W portion of the BIOS 62 followed by BDOS 61 and APPLICATION 59 and COMMAND.COM 60 (APP/COMMAND.COM 59/60). The video memory typically is located above 640K and the ROM BIOS is located in the top-most 64K memory locations below 1M for a 1M address space.
In FIG. 4, the location in memory of the various software components is represented for an alternate embodiment. Again, the interrupt vectors are in the lowest address space. Next is BIOS 62 and APPLICATION 59 and COMMAND.COM 60 (APP/COMMAND.COM 59/60). The video memory is again located above 640K followed by the BDOS 61 including COMMAND.COM. The ROM BIOS is located in the top-most 64K below 1M for a 1M address space. Portions of the BIOS 62 can be stored in ROM.
In FIG. 5, further details of the idle detection system of the computer system of FIG. 1 and FIG. 2 are shown. In FIG. 5, the BDOS 61 including the idle kernel 90 logically connects to the BIOS 62. BIOS 62 includes a plurality of drivers 80 including the drivers 81, 82 . . . , 89 corresponding to the controllers 21, 22, . . . , 29 of FIG. 1. Additionally, BIOS 62 includes an idle driver 79 which in turn may include components 79-1, 79-2, . . . , 79-9 corresponding to the controllers 21, 22, . . . , 29 of the FIG. 1 computer.
A detailed source code listing of the idle kernel 90, IKERNEL.ASM, which is employed in forming the BDOS 61 is given in LISTING 1 at the end of this specification.
A detailed source code listing of the idle driver 79 is given in LISTING 2 through LISTING 6 at the end of this specification. The idle device driver 79 of FIG. 5 is executable in the computer 5 of FIG. 1 in an object code form named IDLE86.SYS. IDLE86.SYS is formed by assembling and linking the source code idle driver IDLE86.SYS of LISTING 2 with the idle custom equates IDLE.EQU of LISTING 3, with the macro equates I:DRMACROS.EQU of LISTING 4, with the request header equates I:REQHDR.EQU of LISTING 5, with the device driver equates I:DRIVER.EQU of LISTING 6.
The object code program, IDLE86.OBJ, is formed by executing the make routine of TABLE 0 as follows.
The idle driver 79 is hardware-dependent code loaded in the form of a character device driver which can be loaded dynamically using the CONFIG.SYS device statement. IDLE86.SYS is loaded, for example, by adding the following statement to the CONFIG.SYS file.
Idle Detection Overview
The idle detection kernel 90 of the BDOS 61 informs the idle driver 79 within the BIOS 62 when BDOS 61 interprets processes in the application program 59 to be idle. A process is idle while it is waiting for some event to occur, for example, an external event such as a keystroke, a mouse movement or a fixed amount of time to pass.
To check for idleness, the idle kernel 90 monitors the operating system (DOS) functions of TABLE 1 which are inactive functions typically used by application programs, such as program 59 of FIG. 2, when they are idle.
IDLE State Data Area
When the idle driver 79 is loaded, DOS opens the driver 79 and, using an IO Control Write, passes the address of a common data area called the IDLE STATE DATA AREA. The device 79 is closed and the driver interface is never used again.
The IDLE STATE DATA AREA is used by both the kernel 90 in DOS and the idle device driver 79 to pass information about the current process state and modify the action of the idle detection code of kernel 90.
The address of this common data area is passed to the driver 79 at the end of the DOS initialization phase. The common data area is used by the idle kernel 90 to communicate process state information to the driver 79 and by the driver 79 to modify the kernel 90 idle detection routines. The process state information which can be passed is set forth in the following TABLE 2.
IDLE.sub.-- COUNT
IDLE.sub.-- COUNT is decremented every time an "IDLE" function call is made by the current DOS process (see table above). All other DOS function calls will reset IDLE.sub.-- COUNT to IDLE.sub.-- MAX. When IDLE.sub.-- COUNT is decremented to 0, the BIOS 61 is called with the PROC-IDLE command.
IDLE.sub.-- MAX
IDLE.sub.-- MAX contains the number of consecutive idle function calls that must be made before the idle driver 79 within the BIOS 62 is called with the PROC.sub.-- IDLE command. This value is initialized to 10 but can be modified by the driver to other values based on the performance of the host computer.
IDLE.sub.-- FLAGS
The idle flags are set forth in the following TABLE 3.
IDLE.sub.-- COMMAND is set whenever the current process is the command processor "COMMAND.COM". This bit is reset when a new application program 59 is loaded or the command processor is terminated. The information for this operation is derived from the history buffer select function.
IDLE.sub.-- DOSFUNC is set whenever an active DOS function call is made. Therefore the BIOS can detect if an application which is polling the keyboard status, using the ROS, is also making any DOS calls. The $IDLE$ driver can reset this bit at any time.
IDLE.sub.-- ON and IDLE.sub.-- ENABLE enable the idle detection code of kernel 90. Only when both flags have been RESET will the idle detection code by enabled. The idle device driver 79 does not modify IDLE.sub.-- ON.
IDLE.sub.-- ENABLE is RESET by idle device driver 79 after IDLE.sub.-- VEC has been initialized in order to enable the kernel 90 idle detection code. This bit can be SET by the device driver 79 if idle detection must be disabled for any reason.
IDLE.sub.-- VEC
This double word pointer is initialized by the idle driver 79 to point to its handler routine which preserves all registers apart from AX and returns using the RETF instruction. The idle device driver 79 is called with the following command codes in the AX register. On entry, the DS register points to the segment of the IDLE State Data Area.
The BIOS IDLE Command Interface is set forth in the following TABLE 4.
PROC.sub.-- IDLE command is executed when the kernel 90 has decided the current process is idling; that is, IDLE.sub.-- MAX consecutive idle DOS function calls have been made. The $IDLE$ device driver 79 may now turn the computer 5 off until the next hardware interrupt occurs.
PROC.sub.-- INT28 is executed when INT28.sub.-- RELOAD consecutive INT28's have been made. The $IDLE$ device driver 79 may now turn the computer off until the next hardware interrupt occurs.
PROC.sub.-- KEYIN is executed when a character is read from the console (CON) device and no character is ready. ES:BX points to the device header for the console. The $IDLE$ driver 79 can now turn the computer off until the next keyboard interrupt unless a TSR's or application is present which may require other interrupts to occur.
PROC.sub.-- DEVIN is executed when a character is read from any device but the console and no character is ready. ES:BX points to the device header. If the device pointed to by ES:BX uses interrupts the $IDLE$ driver 79 can now turn the computer 5 off until a character is sent, unless a TSR's or application program 59 is present which may require other interrupts to occur.
INT28.sub.-- DELAY
When INT28.sub.-- DELAY is decremented to zero, INT28.sub.-- RELOAD consecutive INT28's have been executed and the $IDLE$ device driver is called with the PROC.sub.-- INT28 command.
INT28.sub.-- RELOAD
INT28.sub.-- RELOAD contains the number of consecutive INT28's that must be called before the PROC.sub.-- INT28 command is called.
RESERVED
This double word is reserved as scratch space of the $IDLE$ driver 79. The first word entry is used by example $IDLE$ device driver for the variable IDLE.sub.-- CNTDN.
Device Driver
The device driver 79 contains all the hardware-dependent code to support DOS idle detection. It can either be linked into the BIOS.SYS file of BIOS 62 or loaded dynamically. It supports the four device functions set forth in the following TABLE 5.
The CMD-OUTPUT.sub.-- IOCTL is called to transfer 4 bytes, this is a DWORD pointer to the IDLE State data area. The device driver then initializes the IDLE.sub.-- VEC to point to the handling routine and then resets the IDLE.sub.-- ENABLE bit in IDLE.sub.-- FLAGS.
The $IDLE$ device driver 79 must also monitor any hardware specific status functions that an application program may monitor. Using this information, hardware assistance and the state information, the $IDLE$ device driver 79 turns the system off when the application is not making any DOS calls.
The /V option forces IDLE86.SYS to display real time idle status information on the video monitor. When enabled as a result of an idle sleep function, the first two characters in the top left corner of the screen are incremented to indicate that the idle driver is active. Column 1 row 1 is incremented each time the idle driver is about to put the CPU 10 to sleep to be awakened on any interrupt. Column 2 row 1, is incremented each time the idle driver 79 is about to be put to sleep waiting for a keyboard event. On the right hand side at the top of the screen, a 4 digit hex value is displayed indicating how long it took to come to the decision to go idle. This value is displayed in units of 1/65,353rds of an INT8 timer tick.
The IDLE86.SYS device driver will switch performance states, for example will enter the power saving mode in the embodiment described, in the following circumstances.
1) When the DOS kernel calls through the IDLE.sub.-- VEC.
2) When the IDLE86.SYS device driver finds an application polling the hardware using ROS functions without making any DOS operating system calls.
3) An application reads a character from the keyboard directly using a ROS function.
Before switching performance states by entering the power saving mode, IDLE86.SYS monitors any available information to detect if the application is accessing other components in the system; for example, polling a serial port or updating a graphics screen. If such accessing exists, the device driver returns to the active state immediately and does not switch performance states and does not enter the power saving mode.
The idle driver 79 is able to make the distinction between a program that is genuinely idle, for example one that is polling the keyboard in a tight loop, and one that is active but also polling the keyboard testing for an abort key to be pressed. This distinction is achieved by monitoring the time taken to go idle. If the time is within a specified period as determined by a preestablished parameter, it is assumed that the program is idle, for example polling in a tight loop for a key to be pressed. If the time is outside of the specified limit and therefore differs from the parameter, then it is assumed that some processing has occurred in between polling the keyboard, and the driver should not switch performance states to an idle mode or power-saving state but should return immediately as active state.
It is assumed that a program is idle when an idle state has been detected within the event parameter established by the time period specified in the local variable IDLE.sub.-- CNTDN. The initial value for this event parameter variable is calculated by the following Eq. (1):
where,
INT16.sub.-- TIME is the time taken to perform a ROS INT16h function 1 (keyboard status check).
RTC.sub.-- TIME is the time taken to perform a read of the real-time clock using the ROS INT1Ah function 2 (read real time clock).
Since performance varies from computer to computer, it is not possible to give a static figure for IDLE.sub.-- CNTDN, therefore the value is calculated during execution of the CMD.sub.-- OUTPUT.sub.-- IOCTL function. The times are calculated by reading the 8253 timer chip in the system timer 38 of FIG. 1 at the start and end of the ROS calls, and calculating the elapsed time. Since some TSR keyboard enhancement programs may load and take over the ROS keyboard handler after the $IDLE$ driver 79 has loaded, the $IDLE$ driver dynamically detects if this occurs and recalculates IDLE.sub.-- CNTDN by repeating the test performed during the CMD.sub.-- OUTPUT.sub.-- IOCTL function.
The event parameter established by IDLE.sub.-- CNTDN is compared with the event frequency as determined by the idle kernel 90 and if they differ, the performance state of the computer is switched, that is, the computer goes to sleep into a power-saving mode.
Having decided to switch into a power saving mode, the driver must determine what event will reswitch the system back to a fully active mode. The reswitch condition in general will be any event that is inconsistent with idleness.
If the current application program or a TSR has taken over the timer interrupt in order to "steal" processor time, the system must be "Powered Up" at the next timer interrupt.
Operation With Application Program
As an example of the operation, the computer of FIG. 1 employs the CPU 10, R/W (RAM) memory 13, keyboard 34, real time clock 37 and a CRT display 36 in order to execute a text editor application program represented by FIG. 6. The power saving circuitry in the form of clock unit 6 of FIG. 1, in one example, controls the frequency of the CPU clock, CLK, thereby controlling the rate at which the CPU 10 operates. The clock unit 6 communicates with the idle detection programs to slow down or speed up the CPU clock speed. A slower CPU consumes less power than a fast CPU.
The text editor of FIG. 6 is the application program 59 of FIG. 4 that the computer 5 is running. The operation of the program of FIG. 6 is similar to any word processor application program and serves as a good example for explaining the present invention. When a key on keyboard 34 is pressed, the operating system software translates the electronic signals for that event from the keyboard into a text character that can be displayed onto a computer screen of display 36 or stored in a disk file of disk 31 or 32. Because a typist may enter keystrokes while the computer program 59 is busy performing another task, a "buffer" or portion of the computer 5 memory 13, is set aside to store the incoming keystrokes so that they are ready for use later on.
Thus in the example, when the editor is able to process a keystroke, the buffer is checked to see if a character is available. As the flowchart in FIG. 6 indicates, when a keystroke is available, it is removed from the buffer and inserted into the document text which is also stored in a buffer, and then copied to the screen of display 36 so that the operator can see the character typed. If a keystroke is not made, then the editor effectively has nothing to do, so typically such programs take this opportunity to read the time of day, using the Operating System Software, and display the time on the screen.
The clock 37 in computer 5 has a very fine resolution, that is to say, the clock keeps track of the time of day not just in units of seconds but in fractions of a second. Since the operator is only interested in knowing the time to the nearest second or minute, the time is not redrawn on the screen with the same resolution as the time changes internally in computer 5. Due to the speed at which a computer program executes instructions, the editor loops (waiting for a keystroke event) many hundreds of times before a keystroke event is presented for processing.
With a typist who enters keystrokes at the rate of 180 characters (45 words) a minute, a character is presented every 1/3rd of a second or every 333 msecs (milliseconds). The idle detection system, and particularly the idle kernel 90, detects a possible idle state after IDLE.sub.-- MAX inactive loops of the program by monitoring the number of times that the operating system is called and the type of call that is made. The idle detection driver then checks the display 36 to determine if there was any activity that would suggest that the application program 59 was not idle. In the event that there is no activity, the idle system causes the computer 5 to enter power save mode by either issuing a power save command to the clock unit 6 to reduce the CPU clock speed and/or halt the computer thus reducing the power consumption of the computer 5.
The idle detection driver 79 then resets its internal status information so that a future idle state can be detected and then returns control to the editor application which continues to be executed, all be it at a vastly reduced speed. This reduced speed is quite satisfactory since the program has nothing to do in any case. The idle driver continues to monitor the keyboard and screen waiting for interrupts to occur that would require the CPU to quickly service the event, and therefore, return to full speed and full power consumption.
If suddenly the operator pressed a key that caused an interrupt to be generated, the idle driver detects the interrupt and the CPU is switched back to full power mode automatically. The interrupt is serviced and the keystroke placed into the keyboard buffer, then control is returned to the interrupted editor application. The next time the editor checked the keyboard status as a part of its program loop, it would detect a key-stroke, read the key code from the buffer and process the key code. The editors program loop would then once again check the time to see if it required and then pole the keyboard again waiting for another keystroke and so on and so forth. Eventually, the power save mode is reentered since the idle detection driver again detects inactivity. The CPU clock speed is slowed down and speeded up for short periods of time very often. These short but frequent periods of "rest" or idleness accumulate into considerable power saving even when long periods of idleness do not occur.
Program Execution
Further details of the operation will be apparent in the following LISTING 1, LISTING 2, . . . , LISTING 6.
Through execution of the program routines, the idle detection kernel 90 of LISTING 1 informs the device driver 79 when kernel 90 interprets processes in the application program 59 to be idle. A process is idle while it is waiting for some event to occur, for example, an INT21 inactive event of TABLE 1.
In LISTING 1, the monitoring for the TABLE 1 inactive INT21 events occurs in lines L1-238 through L1-253 in the "int21 func" routine. An inactive event detected in L1-258 through L1-263 will jump to L1-254, L1-255 and to L1-303 to the "inactive" routine. In L1-306, the active.sub.-- cnt value is decremented and tested for zero. If non-zero the process returns and waits for another inactive event detection of the TABLE 1 type. Monitoring for inactive events in this manner continues until the active.sub.-- cnt goes to zero. When zero, the routine in L1-309, L-310 resets the active.sub.-- cnt to idle.sub.-- max parameter and calls the idle driver 79. The call is made in L1-315 by calling the idle.sub.-- vec vector after setting the process idle by setting PROC.sub.-- IDLE in L 1-314.
In L1-258 to L1-260, an active event resets the active count active.sub.-- cnt to idle.sub.-- max.
The idle kernel 90 in monitoring for INT21 inactive events therefore includes monitoring means for monitoring event frequency, by decrementing active.sub.-- cnt, during the execution of the application and other programs by the computer system 5.
The idle kernel 90 does not check every aspect of an application program's operation and therefore the idle driver 79 operates to qualify the idle status information passed by the kernel 90 to the device driver 79. This qualification is done by checking for specific interrupts and using other information about hardware, for example, video monitor operation. In the case of hardware detection with a video monitor, the idle driver 79 reads a data latch to determine if the application is using the video memory. This latch is set whenever the video memory is selected by the CPU.
The idle kernel 90 can call the idle driver in other cases, also. Particularly, when the kernel 90 detects a background application in an INT 28h idle loop, PROC.sub.-- INT28 is set in L1-363 and the idle driver is called in L1-364 using idle.sub.-- vec. When the kernel 90 detects a background application waiting for a keyboard event, PROC.sub.-- KEYIN is set in L1-105 and the idle driver is called in L1-110 using idle.sub.-- vec after a test in L1-106 and jump in L1-107. When the keneral 90 detects a background application waiting for a device input event, PROC.sub.-- DEVIN is set in L1-108 and the idle driver is called in L1-110 using idle.sub.-- vec.
The IDLE86.SYS device driver of LISTING 2 will switch performance states, for example will enter the power saving mode in the embodiment described, in the following circumstances.
1) When the DOS kernel calls through the "idle.sub.-- vec", that is, L1-110, L1-315, and L1-364.
2) When the IDLE86.SYS device driver finds an application polling the hardware using ROS functions without making any DOS operating system calls.
3) An application reads a character from the keyboard directly using a ROS function.
In LISTING 2, the driver is called by the kernel 90 vector accessed from the dword of the data area in L2-165 which calls idle detect in L2-166. In L2-352, the idle.sub.-- detect routine begins and examines for the PROC.sub.-- IDLE, PROC.sub.-- INT28, PROC.sub.-- KEYIN, and PROC.sub.-- DEVIN cases.
Before switching performance states by entering the power saving mode, IDLE86.SYS monitors any available information to detect if the application is accessing other components in the system. For example, THE PROC.sub.-- IDLE case calls the WAIT.sub.-- IDLE routine of L2-536 and checks for the polling of a serial port or updating a graphics screen in L2-537 by calling the routine at L2-628. If such accessing exists, the device driver 79 returns to the active state immediately in L2-634 and does not switch performance states and does not enter the power saving mode.
If no activity is found, the processing returns to the wait.sub.-- idle routine L2-538 for further processing.
The idle driver 79 is able to make the distinction between a program that is genuinely idle, for example one that is polling the keyboard in a tight loop, and one that is active but also polling the keyboard testing for an abort key to be pressed. This distinction is achieved by monitoring the time taken to go idle by calling the chk.sub.-- time routine in L2-543 appearing at L2-574. If the time is within a specified period as determined by a preestablished parameter, it is assumed that the program is idle, for example polling in a tight loop for a key to be pressed. If the time is outside of the specified limit and therefore differs from the parameter, then it is assumed that some processing has occurred in between polling the keyboard, and the driver should not switch performance states to an idle mode or power-saving state but should return immediately as active state.
It is assumed that a program is idle when an idle state has been detected within the event parameter established by the time period specified in the local variable IDLE.sub.-- CNTDN. The value for this event parameter variable is calculated by the calc.sub.-- idlecntdn routine at line L2-941.
The event parameter established by IDLE.sub.-- CNTDN is compared with the event frequency as determined by the idle kernel 90 in L2-599 and if not exceeded, the performance state of the computer is switched, that is, the computer goes to sleep into a power-saving mode by the jump at L2-545 in the wait.sub.-- idle routine. If the time to idle has been too long, the return from chk.sub.-- time at L2-606 will cause the jump on condition at L2-544 to bypass L2-545 and hence will not jump to go.sub.-- to.sub.-- sleep.
Having decided to switch into a power saving mode, the driver must determine what event will reswitch the system back to a fully active mode. The reswitch condition in general will be any event that is inconsistent with idleness.
The foregoing and other objects, features and advantages of the invention will be further apparent from the following detailed embodiment setforth in TABLE 1, TABLE 2, . . . , TABLE 6. ##SPC1##
While the invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that the foregoing and other changes in form and details may be made therein without departing from the spirit and scope of the invention.
Claims
What is claimed is:
1. An activity system in a computer where the computer executes programs involving multiple events with times between events and where the computer operates with multiple performance states comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency represented by an event frequency value proportional to the frequency of said plurality of said multiple events during the execution of the programs by the computer system, parameter means for establishing an event parameter, comparing means for comparing the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching means for switching the performance state of the computer when the event frequency differs from the event parameter.
2. The activity system of claim 1 wherein said switching means includes reswitch means for switching the performance state of the computer in response to a reswitch condition.
3. The activity system of claim 2 wherein said reswitch means switches the performance state of the computer in response to a time-out reswitch condition.
4. The activity system of claim 2 wherein said reswitch means switches the performance state of the computer in response to an external reswitch condition.
5. The activity system of claim 2 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit and wherein said reswitch means switches the performance state of the computer in response to an interrupt irrespective of when the switching means switched the performance state.
6. The activity system of claim 1 wherein said computer performance states include a high-power state and a low-power state and said switching means switches to the low-power state when the event frequency differs from the event parameter.
7. The activity system of claim 1 wherein said computer performance states include a high-frequency state and a low-frequency state and said switching means switches to one of said states when the event frequency differs from the event parameter.
8. The activity system of claim 1 wherein said multiple events originate externally from said computer system whereby said event frequency is independent of the performance of said computer system.
9. The activity system of claim 8 wherein said multiple events originate from user input at input devices for the computer.
10. The activity system of claim 9 wherein said multiple events originate from user keystroke input at a keyboard device for the computer.
11. The activity system of claim 1 wherein, said monitoring means monitors the actual time between events as said event frequency, said parameter means establishes said event parameter as a maximum event time to elapse between events, said comparing means compares the actual time with the event time to determine when the event frequency differs from the event parameter by determining when the actual time is less than the event time.
12. The activity system of claim 11 wherein, said parameter means includes event setting means to set said event time as a variable as a function of the performance of the computer system.
13. The activity system of claim 11 wherein, said parameter means includes event setting means to set said event time as a variable as a function of inputs external to said computer system.
14. An activity system in a computer where the computer executes programs involving multiple events with times between events and where the computer operates with multiple performance states, wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency during the execution of the programs by the computer system, said monitoring means including, interrupt timer means for monitoring an interrupt time as the time taken to perform an interrupt in the computer system, clock timer means for monitoring a clock time as the time taken to perform a read of the real-time clock, parameter means for establishing an event parameter as an event time, and including means for setting said event time as a function of the sum of the clock time and the interrupt time, comparing means for comparing the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching means for switching the performance state of the computer when the event frequency differs from the event parameter.
15. The activity system of claim 14 wherein said computer includes memory means for storing kernel code for said monitoring means in a kernel location and memory means for storing driver code for said parameter means in a driver location.
16. The activity system of claim 15 wherein said computer includes memory means for storing state information for use by said monitoring means and said parameter means for transferring information between said kernel code and said driver code.
17. An activity system in a computer system where the computer system includes an operating system for controlling operations of the computer when the computer executes programs involving multiple events with times between events and where the computer operates with multiple performance states comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency represented by an event frequency value proportional to the frequency of said plurality of said multiple events during the execution of the programs by the computer system, parameter means under control of the operating system for establishing an event parameter, comparing means under control of the operating system for comparing the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching means for switching the performance state of the computer when the event frequency differs from the event parameter, reswitch means for switching the performance state of the computer in response to a reswitch condition.
18. The activity system of claim 17 wherein said computer includes memory means for storing said operating system and wherein said operating system includes kernel code for said monitoring means and driver code for said parameter means.
19. The activity system of claim 18 wherein said computer includes memory means for storing state information for use by said kernel code and said driver code for transferring information between said monitoring means and said parameter means.
20. The activity system of claim 17 wherein, said monitoring means monitors the actual time between events as said event frequency, said parameter means establishes said event parameter as a maximum event time to elapse between events, said comparing means compares the actual time with the event time to determine when the event frequency differs from the event parameter by determining when the actual time is less than the event time.
21. The activity system of claim 20 wherein, said parameter means includes event setting means to set said event time as a variable as a function of the performance of the computer system.
22. The activity system of claim 21 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock and wherein, said monitoring means includes, interrupt timer means for monitoring an interrupt time as the time taken to perform an interrupt in the computer system, clock timer means for monitoring a clock time as the time taken to perform a read of the real-time clock, said event setting means includes, means for setting said event time as a function of the sum of the clock time and the interrupt time.
23. The activity system of claim 17 including means for counting system time at different rates for different performance states and wherein said computer includes a real time clock for counting real time, said activity system including means for updating said system time with said real time as a function of switching performance states.
24. The activity system of claim 17 including means for specifying said reswitch condition.
25. The activity system of claim 17 wherein said computer includes a keyboard having keyboard activity and wherein said reswitch means includes means for specifying said reswitch condition as a function of keyboard activity.
26. The activity system of claim 17 wherein said computer includes a real time clock for establishing real time and wherein said reswitch means includes means for specifying said reswitch condition as a function of a real time value.
27. The activity system of claim 17 wherein said computer system is adapted to execute an application program and wherein said reswitch means includes means for specifying said reswitch condition as a function of a command from said application program.
28. An activity system in a computer system where the computer system includes an operating system for controlling operations of the computer when the computer executes programs involving multiple events with times between events and where the computer operates with a low performance state and a high performance state comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency represented by an event frequency value proportional to the frequency of said plurality of said multiple events during the execution of the programs by the computer system, parameter means under control of the operating system for establishing an event parameter, comparing means under control of the operating system for comparing the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching means for switching the performance state of the computer from the high performance state to the low performance state when the event frequency differs from the event parameter, reswitch means for switching the performance of the computer from the low performance state to the high performance state in response to a reswitch condition.
29. The activity system of claim 28 including reswitch means for setting said reswitch condition.
30. The activity system of claim 29 wherein said reswitch means includes means for setting said reswitch condition as a function of multiple conditions.
31. The activity system of claim 30 wherein said computer includes a central processing unit performing functions in response to multiple types of interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes means for selecting types of interrupts as said multiple conditions.
32. The activity system of claim 28 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock and wherein, said monitoring means includes, interrupt timer means for monitoring an interrupt time as the time taken to perform an interrupt in the computer system, clock timer means for monitoring a clock time as the time taken to perform a read of the real-time clock, said event setting means includes, means for setting said event time as a function of the sum of the clock time and the interrupt time.
33. An activity system method in a computer where the computer executes programs involving multiple events with times between events and where the computer operates with multiple performance states comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency represented by an event frequency value proportional to the frequency of said plurality of said multiple events during the execution of the programs by the computer system, establishing an event parameter, comparing the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching the performance state of the computer when the event frequency differs from the event parameter.
34. The activity system method of claim 33 wherein said switching step includes a reswitch step for switching the performance state of the computer in response to a reswitch condition.
35. The activity system method of claim 34 wherein said reswitch step switches the performance state of the computer in response to a time-out reswitch condition.
36. The activity system method of claim 34 wherein said reswitch step switches the performance state of the computer in response to an external reswitch condition.
37. The activity system method of claim 34 wherein said computer performs functions a central processing unit in response to interrupts, generates interrupts to said central processing unit and wherein said reswitch step switches the performance state of the computer in response to an interrupt.
38. The activity system method of claim 33 wherein said computer performance states include a high-power state and a low-power state and said switching step switches to the low-power state when the event frequency differs from the event parameter.
39. The activity system method of claim 33 wherein said computer performance states include a high-frequency state and a low-frequency state and said switching step switches to one of said states when the event frequency differs from the event parameter.
40. The activity system method of claim 33 wherein said multiple events originate externally from said computer system whereby said event frequency is independent of the performance of said computer system.
41. The activity system method of claim 40 wherein said multiple events originate from user input at input devices for the computer.
42. The activity system method of claim 41 wherein said multiple events originate from user keystroke input at a keyboard device for the computer.
43. The activity system method of claim 33 wherein, said monitoring step monitors the actual time between events as said event frequency, said parameter step establishes said event parameter as a maximum event time to elapse between events, said comparing step compares the actual time with the event time to determine when the event frequency differs from the event parameter by determining when the actual time is less than the event time.
44. The activity system method of claim 43 wherein, said parameter step includes an event setting step to set said event time as a variable as a function of the performance of the computer system.
45. The activity system method of claim 43 wherein, said parameter step includes an event setting step to set said event time as a variable as a function of inputs external to said computer system.
46. The activity system method of claim 33 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock and wherein, said monitoring step includes, an interrupt timer step for monitoring an interrupt time as the time taken to perform an interrupt in the computer system, a clock timer step for monitoring a clock time as the time taken to perform a read of the real-time clock, said event setting step includes, setting said event time as a function of the sum of the clock time and the interrupt time.
47. The activity system method of claim 33 wherein said computer includes memory for storing kernel code for said monitoring step in a kernel location and memory for storing driver code for said parameter step in a driver location.
48. The activity system method of claim 47 wherein said computer includes memory for storing state information for use by said monitoring step and said parameter step for transferring information between said kernel code and said driver code.
49. An activity system method in a computer system where the computer system includes an operating system for controlling operations of the computer when the computer executes programs involving multiple events with times between events and where the computer operates with multiple performance states comprising, monitoring under control of the operating system a plurality of said multiple events to determine event frequency during the execution of the programs by the computer system, establishing under control of the operating system an event parameter, comparing under control of the operating system the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching the performance state of the computer when the event frequency differs from the event parameter, reswitching the performance state of the computer in response to a reswitch condition.
50. The activity system method of claim 49 wherein said computer includes memory step for storing said operating system and wherein said operating system includes kernel code for said monitoring step and driver code for said parameter step.
51. The activity system method of claim 50 wherein said computer includes memory for storing state information for use by said kernel code and said driver code for transferring information between said monitoring step and said parameter step.
52. The activity system method of claim 49 wherein, said monitoring step monitors the actual time between events as said event frequency, said parameter step establishes said event parameter as a maximum event time to elapse between events, said comparing step compares the actual time with the event time to determine when the event frequency differs from the event parameter by determining when the actual time is less than the event time.
53. The activity system method of claim 52 wherein, said parameter step includes event setting step to set said event time as a variable as a function of the performance of the computer system.
54. The activity system method of claim 53 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock and wherein, said monitoring step includes, monitoring an interrupt time as the time taken to perform an interrupt in the computer system, monitoring a clock time as the time taken to perform a read of the real-time clock, said event setting step includes, setting said event time as a function of the sum of the clock time and the interrupt time.
55. The activity system method of claim 49 including counting system time at different rates for different performance states and wherein said computer includes a real time clock for counting real time, said activity system method including the step of updating said system time with said real time as a function of switching performance states.
56. The activity system method of claim 49 including the step for specifying said reswitch condition.
57. The activity system method of claim 49 wherein said computer includes a keyboard adapted for having keyboard activity and wherein said reswitch step includes the step of specifying said reswitch condition as a function of keyboard activity.
58. The activity system method of claim 49 wherein said computer includes a real time clock for establishing real time and wherein said reswitch step includes the step of specifying said reswitch condition as a function of a real time value.
59. The activity system method of claim 49 wherein said computer system is adapted to execute an application program and wherein said reswitch step includes the step of specifying said reswitch condition as a function of a command from said application program.
60. An activity system method in a computer system where the computer system includes an operating system for controlling operations of the computer when the computer executes programs involving multiple events with times between events and where the computer operates with a low performance state and a high performance state comprising, monitoring means for monitoring a plurality of said multiple events to determine event frequency represented by an event frequency value proportional to the frequency of said plurality of said multiple events during the execution of the programs by the computer system, establishing under control of the operating system an event parameter, comparing under control of the operating system the event frequency with the event parameter to determine when the event frequency differs from the event parameter, switching the performance state of the computer from the high performance state to the low performance state when the event frequency differs from the event parameter, reswitching the performance state of the computer from the low performance state to the high performance state in response to a reswitch condition.
61. The activity system method of claim 60 including a reswitch step for setting said reswitch condition.
62. The activity system method of claim 61 wherein said reswitch step includes the step of setting said reswitch condition as a function of multiple conditions.
63. The activity system method of claim 62 wherein said computer includes a central processing unit performing functions in response to multiple types of interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes the step of selecting types of interrupts as said multiple conditions.
64. The activity system method of claim 60 wherein said computer includes a central processing unit performing functions in response to interrupts, includes interrupt generating means for generating interrupts to said central processing unit, and includes a real time clock and wherein, said monitoring step includes, monitoring an interrupt time as the time taken to perform an interrupt in the computer system, monitoring a clock time as the time taken to perform a read of the real-time clock, said event setting step includes, setting said event time as a function of the sum of the clock time and the interrupt time.
65. An activity system method in a computer system where the computer system includes an operating system for controlling operations of the computer when the computer executes programs involving multiple operating system events with times between events and where the computer operates with a low performance state and a high performance state comprising, monitoring means for monitoring a plurality of said multiple operating system events to determine event frequency of said multiple events during the execution of the programs by the computer system, establishing under control of the operating system an event parameter, comparing under control of the operating system the event frequency with the event parameter to determine when the event frequency differs form the event parameter, switching the performance state of the computer from the high performance state to the low performance state when the event frequency differs from the event parameter, reswitching the performance state of the computer form the low performance state to the high performance state in response to a reswitch condition.
Patent Citations (7)
Non-Patent Literature (2)
- WO 91/00566, PCT International Application, published Jan. 10, 1991, Poquet Computer Corp.
- 0364222, European Patent Application, published Apr. 18, 1990, Compaq Computer Corp.