Selective power-down for high performance CPU/system

Details Selective power-down for high performance CPU/system Selective power-down for high performance CPU/system

2001-05-10

2002-08-06

Dharia, Rupal (Department: 2181)

Electrical computers and digital processing systems: support

Computer power control

Power conservation

C713S310000, C713S321000, C713S324000

Reexamination Certificate

active

06430693

ABSTRACT:

The following are commonly owned, co-pending applications:
“Superscalar RISC Instruction Scheduling”, application Ser. No. 08/219,425 (now U.S. Pat. No. 5,497,499); and
“Hardware Emulation Accelerator and Method”, application Ser. No. 08/352,680 (now U.S. Pat. No. 5,581,742).
The disclosures of the above applications are incorporated herein by reference.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to systems and methods for reducing power dissipation requirements and power consumed by single microelectronic devices, and more particularly, to dynamic control of power consumption by and resultant power dissipation required of such microelectronic devices.
2. Background Art
Power dissipation requirements of microelectronic devices (also called semiconductor devices or semiconductor chips or integrated circuits) have become critical in their design, fabrication and use. This is particularly true with very large scale integrated (VLSI) devices and ultra large scale integrated (ULSI) devices, which typically today have over 1,000,000 transistors (active passive) on a single semiconductor die. The active devices are typically run (clocked) at very high speed (25 MHz and 33 MHz speeds are now typical, with much higher clock rates contemplated, for example, 250 MHz) in order to achieve desired system functionality and performance.
As is well known, the high clock rate and the high number of active devices, regardless of the fabrication technology that is used, produce significant power dissipation requirements when compared to the actual physical size of the die of the microelectronic device. For purposes of illustration, a typical die with 1,000,000 active devices is fabricated on a die 15 mm by 15 mm and requires greater than 132 pinouts. Such a microelectronic device can operate at a system clock speed of 30 MHz with 1 micrometer (&mgr;m) CMOS technology.
The die must be permanently housed in a suitable housing or package, which among other things (pin out, environmental, physical protection, etc.) must provide adequate heat dissipation in order to prevent failure of the device.
It is not uncommon for a single microelectronic device such as the example above to generate in the range of 5-10 watts of heat that must be dissipated during normal operation. As a result, the junction temperature of the die of such a microelectronic device can reach 100° C. for a ceramic package without heatsinking, at the high end of the commercial temperature range, 70° ambient. The 5-10 watt number will look small compared to the dissipation requirement for successive generations of more powerful microelectronic devices, which are projected by year 2000 to have 100 million active devices on a single die. Contemplated die sizes are 25 mm by 25 mm.
Various strategies for packaging have evolved to deal with large heat dissipation. All include some type of heat sink or thermal grease arrangement for rapidly drawing away the unwanted heat so as to protect the microelectronic device (die and bond wires) from physical failure and performance degradation. Gas, such as air, and even liquid, such as water, freon, and more efficient coolants are typically used in addition to a conventional heat sink. Heat sink approaches, however, act to increase physical size, cost, mechanical complexity, and weight of the packaged microelectronic device. Moreover, the heat dissipation (i.e., thermal stress) requirements act to limit the physical size of a die that can be accommodated in a single package.
Representative of the heat dissipation requirements are conventional microcrocessor chips running at clock speeds up to 50 MHz, which can typically generate 5 watts of dissipated power in normal operation. In order to accommodate the heat dissipation requirement, special heat sinks arrangement are provided.
The constant trend in electronics is to reduce the size of microelectronic devices so that smaller and lighter electronic and computer products can be made. This miniaturization drive goes on unabated, and historically produces from year to year dramatic reductions in physical size.
The heat dissipation requirement, however, acts as a barrier to this miniaturization process of electronic and computer devices. In other words, the physics of having to dissipate the heat from the microelectronic device limits the physical size and weight reduction of the electronic or computer device that can be achieved. This affects the lifetime of microelectronic devices as well. For example, the reason why a solid state laser has a shorter lifetime than an LED is due to concentration of heat at a small area.
Another significant ongoing trend in electronics is the increase in the features and functions and the decrease in response time that can be provided by an electronic or computer device. This is achieved through more complex and powerful microelectronic devices. This is the result of the increased integration of active devices on a single die. However, additional active devices on the die results in increased heat dissipation requirements, which acts to limit the reduction in the size of the microelectronic device package that can be achieved. Even by reducing the power supply voltage, DEC's Alpha CMOS chip, for example, is reported to dissipate 30 watts at 200 Mhz.
The dramatic decrease in the physical size of microelectronic devices when compared to their computational capability, and features and functions that they can produce, has resulted in the creation of very small personal computers, typically called laptop, notebook and palmtop computers. This is the latest benchmark in an ongoing trend to reduce in size computers having powerful features and functions.
A typical portable computer today having a 386SX type microprocessor has physical dimensions of 12 in. by 16 in., and a weight of 15 lbs., of which 1 lb. is the rechargeable battery. A typical laptop computer today having a 386SXL type microprocessor has physical dimensions of 8 in. by 11 in. by 2 in. and a weight of 5-7 lbs., of which 0.5 lbs. is the rechargeable battery.
One of the most critical limiting factors, however, to such notebook (also laptop and palmtop) computers is the battery that is needed to run the machine. The battery must provide sufficient electrical power so that the computer can operate for a long enough period of time to satisfy user demand. Typical operating time for notebook computers today is in the range of 3 to 4 hours for a single battery charge.
The battery comprises one of the largest components of the computer system in terms of weight and physical size. However, it is critical for the user that enough electrical power be provided by the battery so that desired computer operation can occur over a sufficient period of time. However, this requirement for operability causes the total size of the computer system to increase since the battery physical size must be increased to meet these requirements.
Consequently, considerable research and development is being directed towards producing much more efficient batteries for a given size and weight. The goal here is to increase battery technology in charge capacity so that the resultant battery will provide more power and longer time for the given size and space. This will in turn act to reduce the size of the computer system that uses it.
In addition to reducing the size of the battery, considerable effort is being expended to try to increase the performance of the computer system in terms of power consumption. One conventional approach as utilized by Intel is to turn off unused peripheral chips. This occurs in the Intel 80386 chip set. By turning off unused peripheral chips, significant battery life can be achieved because the peripheral chips consumed considerable amounts of power.
A further approach implemented in AMD's AM386DXL microprocessor chip is to slow down the clock speed (e.g., from 40-0 MHz) to conserve power.
In view of the above, there is a great need for improvement in heat dissipation and power consumption by microelectronic d

Affiliated with

Also associated with

Rating

Selective power-down for high performance CPU/system does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Selective power-down for high performance CPU/system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Selective power-down for high performance CPU/system will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2911128