U.S. patent application number 10/142187 was filed with the patent office on 2003-11-13 for technique for optimizing decoupling capacitance subject to leakage power constraints.
Invention is credited to Bobba, Sudhakar, Thorp, Tyler, Trivedi, Pradeep.
Application Number | 20030212965 10/142187 |
Document ID | / |
Family ID | 29399825 |
Filed Date | 2003-11-13 |
United States Patent
Application |
20030212965 |
Kind Code |
A1 |
Bobba, Sudhakar ; et
al. |
November 13, 2003 |
TECHNIQUE FOR OPTIMIZING DECOUPLING CAPACITANCE SUBJECT TO LEAKAGE
POWER CONSTRAINTS
Abstract
A technique for optimizing decoupling capacitance on an
integrated circuit while meeting leakage power constraints of the
integrated circuit is provided. The technique involves the
formulation of a linear optimization problem using physical
characteristics and constraints of the integrated circuit, where a
linear solution to the linear optimization problem yields an
optimal decoupling capacitance presence on the integrated
circuit.
Inventors: |
Bobba, Sudhakar; (Sunnyvale,
CA) ; Trivedi, Pradeep; (Sunnyvale, CA) ;
Thorp, Tyler; (Sunnyvale, CA) |
Correspondence
Address: |
ROSENTHAL & OSHA L.L.P. / SUN
1221 MCKINNEY, SUITE 2800
HOUSTON
TX
77010
US
|
Family ID: |
29399825 |
Appl. No.: |
10/142187 |
Filed: |
May 9, 2002 |
Current U.S.
Class: |
716/115 |
Current CPC
Class: |
G06F 30/36 20200101;
G06F 30/367 20200101 |
Class at
Publication: |
716/2 |
International
Class: |
G06F 017/50 |
Claims
What is claimed is:
1. A method for assigning thin-oxide and thick-oxide capacitors on
an integrated circuit, the integrated circuit having a capacitance
requirement and a leakage power constraint, the method comprising:
formulating a linear optimization problem, wherein formulating the
linear optimization problem comprises: defining a first value as a
value of capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors subject to the leakage
power constraint; solving the linear optimization problem to
determine a maximum value for the first value; and determining an
optimal percentage of thin-oxide capacitance using the maximum
value.
2. The method of claim 1, wherein the leakage power constraint is
for a particular region on the integrated circuit, and wherein the
optimal percentage of thin-oxide capacitance is determined for the
particular region.
3. The method of claim 1, wherein the determination of the maximum
value of capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors is dependent on an area
of the integrated circuit.
4. The method of claim 1, wherein the maximum value of capacitance
that is obtained by replacing the thick-oxide capacitors with the
thin-oxide capacitors is determined over a plurality of regions on
the integrated circuit.
5. The method of claim 4, wherein the determination of the maximum
value of capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors is dependent on areas of
the plurality of regions.
6. A computer system, comprising: a processor; a memory; and
instructions, residing in the memory and executable by the
processor, for: formulating a linear optimization problem, wherein
formulating the linear optimization problem comprises: defining a
first value as a value of capacitance that is obtained by replacing
the thick-oxide capacitors with the thin-oxide capacitors subject
to the leakage power constraint; solving the linear optimization
problem to determine a maximum value for the first value; and
determining an optimal percentage of thin-oxide capacitance using
the maximum value.
7. The computer system of claim 6, wherein the leakage power
constraint is for a particular region on the integrated circuit,
and wherein the optimal percentage of thin-oxide capacitance is
determined for the particular region.
8. The computer system of claim 6, wherein the determination of the
maximum value of capacitance that is obtained by replacing the
thick-oxide capacitors with the thin-oxide capacitors is dependent
on an area of the integrated circuit.
9. The computer system of claim 6, wherein the maximum value of
capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors is determined over a
plurality of regions on the integrated circuit.
10. The computer system of claim 9, wherein the determination of
the maximum value of capacitance that is obtained by replacing the
thick-oxide capacitors with the thin-oxide capacitors is dependent
on areas of the plurality of regions.
11. A computer-readable medium having recorded therein instructions
executable by processing, the instructions for: formulating a
linear optimization problem, wherein formulating the linear
optimization problem comprises: defining a first value as a value
of capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors subject to the leakage
power constraint; and determining an optimal percentage of
thin-oxide capacitance dependent on a maximization of the first
value.
12. The computer system of claim 11, wherein the leakage power
constraint is for a particular region on the integrated circuit,
and wherein the optimal percentage of thin-oxide capacitance is
determined for the particular region.
13. The computer system of claim 11, wherein the determination of
the maximum value of capacitance that is obtained by replacing the
thick-oxide capacitors with the thin-oxide capacitors is dependent
on an area of the integrated circuit.
14. The computer system of claim 11, wherein the maximum value of
capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors is determined over a
plurality of regions on the integrated circuit.
15. The computer system of claim 15, wherein the determination of
the maximum value of capacitance that is obtained by replacing the
thick-oxide capacitors with the thin-oxide capacitors is dependent
on areas of the plurality of regions.
16. A method for assigning thin-oxide and thick-oxide capacitors on
an integrated circuit, the integrated circuit having a capacitance
requirement and a leakage power constraint, the method comprising:
step for formulating a linear optimization problem, wherein the
step for formulating the linear optimization problem comprises:
defining a first value as a value of capacitance that is obtained
by replacing the thick-oxide capacitors with the thin-oxide
capacitors subject to the leakage power constraint; step for
solving the linear optimization problem to determine a maximum
value for the first value; and step for determining an optimal
percentage of thin-oxide capacitance using the maximum value.
Description
BACKGROUND OF INVENTION
[0001] A typical computer system has at least a microprocessor and
memory. The microprocessor processes, i.e., executes, instructions
to accomplish various tasks of the computer system. Such
instructions, along with the data required by the microprocessor
when executing these instructions, are stored in some form of
memory. FIG. 1 shows a typical computer system having a
microprocessor (10) and some form of memory (12). The
microprocessor (10) has, among other components, a central
processing unit (also known and referred to as "CPU" or "execution
unit") (14) and a memory controller (also known as "load/store
unit") (16). The CPU (14) is where the actual arithmetic and
logical operations of the computer system take place. To facilitate
the execution of operations by the CPU (14), the memory controller
(16) provides the CPU (14) with necessary instructions and data
from the memory (12). The memory controller (16) also stores
information generated by the CPU (14) into the memory (12).
[0002] The operations that occur in a computer system, such as the
logical operations in the CPU and the transfer of data between the
CPU and memory, require power. If the components responsible for
carrying out specific operations do not receive adequate power in a
timely manner, computer system performance is susceptible to
degradation. As an added challenge, power consumption of modern
computers has increased as a consequence of increased operating
frequencies. Thus, providing power to the components in a computer
system in a sufficient and timely manner has become an issue of
significant importance.
[0003] Often, power supply to a particular computer system element
varies, which, in turn, effects the integrity of the element's
output. Typically, this power variation results from the distance
between a power supply for the element and the element itself. This
distance may lead to the element not receiving power (via current)
at the exact time it is required.
[0004] As shown in FIG. 2, one approach used by designers to combat
this performance-inhibiting behavior is introducing decoupling
capacitance to a particular circuit by positioning one or more
decoupling capacitors (13) close to elements (15) in an integrated
circuit (17). These decoupling capacitors (13) store charge from
the power supply and distribute the charge to the elements (15)
when needed. For example, if power received by a element from a
power supply line (19) attenuates, one or more decoupling
capacitors (13) will distribute charge to the element (15) to
ensure that the element (15) is not affected by the power variation
on the power supply line (19). In essence, a decoupling capacitor
acts as a local power supply for one or more specific elements in a
computer system.
[0005] However, important considerations must be made as to the
assignment of one or more decoupling capacitors to particular
capacitance needing elements because capacitors have particular
undesirable characteristics. One such characteristic pertains to
two types of capacitors: thin-oxide capacitors and thick-oxide
capacitors. A thin-oxide capacitor is designed using one or more
transistors that have thin gate dielectric thicknesses, and
although thin-oxide capacitors provide a relatively large amount of
decoupling capacitance, they are prone to undesirable
gate-tunneling leakage currents. Such leakage current, in turn,
increases the leakage power of a circuit, resulting in increased
power and heat dissipation by the circuit. Alternatively, a
thick-oxide capacitor is designed using one or more transistors
that have thick gate dielectric thicknesses, and although
thick-oxide capacitors have less leakage currents, they provide a
smaller amount of decoupling capacitance than thin-oxide
capacitors. Thus, there is a need for a technique that optimizes
decoupling capacitance such that decoupling capacitance on an
integrated circuit is increased while leakage power constraints are
met.
SUMMARY OF INVENTION
[0006] According to one aspect of the present invention, a method
for assigning thin-oxide and thick-oxide capacitors on an
integrated circuit, where the integrated circuit has a capacitance
requirement and a leakage power constraint, comprises: formulating
a linear optimization problem, where formulating the linear
optimization problem comprises defining a first value as a value of
capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors subject to the leakage
power constraint; solving the linear optimization problem to
determine a maximum value for the first value; and determining an
optimal percentage of thin-oxide capacitance using the maximum
value.
[0007] According to another aspect, a computer system comprises a
processor, a memory, and instructions, residing in the memory and
executable by the processor, for: formulating a linear optimization
problem, where formulating the linear optimization problem
comprises defining a first value as a value of capacitance that is
obtained by replacing the thick-oxide capacitors with the
thin-oxide capacitors subject to the leakage power constraint;
solving the linear optimization problem to determine a maximum
value for the first value; and determining an optimal percentage of
thin-oxide capacitance using the maximum value.
[0008] According to another aspect, a computer-readable medium
having recorded therein instructions executable by processing
comprises instructions for: formulating a linear optimization
problem, where formulating the linear optimization problem
comprises defining a first value as a value of capacitance that is
obtained by replacing the thick-oxide capacitors with the
thin-oxide capacitors subject to the leakage power constraint; and
determining an optimal percentage of thin-oxide capacitance
dependent on a maximization of the first value.
[0009] According to another aspect, a method for assigning
thin-oxide and thick-oxide capacitors on an integrated circuit,
where the integrated circuit has a capacitance requirement and a
leakage power constraint, comprises: a step for formulating a
linear optimization problem, where the step for formulating the
linear optimization problem comprises defining a first value as a
value of capacitance that is obtained by replacing the thick-oxide
capacitors with the thin-oxide capacitors subject to the leakage
power constraint; a step for solving the linear optimization
problem to determine a maximum value for the first value; and a
step for determining an optimal percentage of thin-oxide
capacitance using the maximum value.
[0010] Other aspects and advantages of the invention will be
apparent from the following description and the appended
claims.
BRIEF DESCRIPTION OF DRAWINGS
[0011] FIG. 1 shows a typical computer system.
[0012] FIG. 2 shows a typical arrangement of decoupling capacitors
and circuit elements.
[0013] FIG. 3a shows an integrated circuit that is referenced by
the discussion of FIG. 3b.
[0014] FIG. 3b shows a flow process in accordance with an
embodiment of the present invention.
[0015] FIG. 4 shows a computer system in accordance with an
embodiment of the present invention.
DETAILED DESCRIPTION
[0016] Embodiments of the present invention relate to a technique
for optimizing decoupling capacitance on an integrated circuit
while meeting leakage power constraints. Embodiments of the present
invention further relate to a technique for formulating a linear
optimization problem for optimizing decoupling capacitance on an
integrated circuit, where the linear optimization problem is
solvable using a linear solver program.
[0017] FIG. 3a shows an integrated circuit (20) that will be used
as a reference for the discussion of the present invention. The
integrated circuit (20) is made up of several regions 1 . . . n,
where n represents the number of regions being considered. It is
desired to optimize the amount of capacitance in each of these
regions while meeting leakage power budgets of either the entire
integrated circuit or each of the regions shown in FIG. 3a.
[0018] Determining an optimal decoupling capacitance arrangement
involves determining a maximum value of the following term (1)
subject to the constraint equation (2): 1 k = 1 n a k x k , ( 1
)
[0019] where a.sub.k represents the extra capacitance obtained by
replacing thick-oxide capacitors with thin-oxide capacitors in the
kth region, and where x.sub.k represents the percentage of
thin-oxide decoupling capacitance in the kth region.
[0020] Those skilled in the art will understand that because x
represents a percentage, the value of x is between 0 and 1,
inclusive. Those skilled in the art will also understand that
a.sub.k is normalized to the area of the kth region. Further, those
skilled in the art will appreciate that x can be a discrete value
between 0 and 1, inclusive.
[0021] As stated above, the determination of the maximum value for
term (1) is subject to the following constraint equation (2): 2 k =
1 n w k x k C , ( 2 )
[0022] where w.sub.k represents the extra leakage current resulting
from the replacement of thick-oxide capacitors by thin-oxide
capacitors in the kth region, and where C represents the leakage
power budget for the integrated circuit (20).
[0023] In alternative embodiments, a leakage power constraint may
be particular to a region instead of the entire integrated circuit
(20), in which case the constraint equation for that particular
region is:
w.sub.ix.sub.i<C.sub.i, (3)
[0024] where w.sub.i represents the extra leakage current resulting
from the replacement of thick-oxide capacitors by thin-oxide
capacitors in that particular ith region, where x.sub.i represents
the percentage of thin-oxide capacitors in the ith region, and
where C.sub.i represents the leakage power budget for the ith
region.
[0025] Equations/terms (1), (2), and (3) above form a linear
optimization problem with linear constraints, and therefore, by
solving for the maximum value of term (1) subject to equation (2)
and/or possibly equation (3), a value for x may be determined,
where x represents what percentage of thin-oxide capacitance that
can be used in particular regions of the integrated circuit or in
the integrated circuit as a whole. More particularly, the solved
value of x represents an optimal amount of thin-oxide capacitance
that can be used without violating leakage power constraints.
[0026] Those skilled in the art will appreciate that the linear
optimization algorithm developed above may be solved using any
number of linear problem solving techniques. For example, one or
ordinary skill in the art will understand the aforementioned linear
optimization problem may be solved using LaGrange multipliers. In
another example, linear programming software may be used to
determine an optimal amount of thin-oxide capacitance given the
formulation technique above.
[0027] FIG. 3b shows a flow process describing a technique for
optimizing decoupling capacitance on the integrated circuit (20) in
accordance with an embodiment of the present invention. Initially,
a determination is made as to the percentage of thin-oxide
capacitance based on the extra leakage current resulting in a
region when the thick-oxide capacitance is replaced by the
thin-oxide capacitance in the region (step 22). This may be
repeated for a desired number of regions (step 24). Thereafter, a
determination is made as to the optimal value of capacitance based
on the percentage of thin-oxide capacitance determined above and
the capacitance resulting from the replacement of the thick-oxide
with the thin-oxide capacitance (step 26). This determination of
the optimal value may be made based on a plurality of regions on
the integrated circuit (20) (step 28).
[0028] FIG. 4 shows an exemplary computer system (30) that
determines an optimal decoupling capacitance in accordance with an
embodiment of the present invention. Input parameters (32) may
include a circuit design or layout, an available capacitance area
size for a particular area being considered, an amount of
thin-oxide capacitors available for the area being considered, and
leakage power budget information for the integrated circuit or
particular regions thereon. One of ordinary skill in the art will
understand that the input parameters (32) may include additional
values, such as information relating to per unit areas of
thin-oxide and thick-oxide capacitances present on the integrated
circuit or particular regions thereon.
[0029] The input parameters (32) serve as input data to the
computer system (30) via some computer-readable medium, e.g.,
network path, floppy disk, input file, etc. The computer system
(30) then stores the input parameters (32) in memory (not shown) to
subsequently determine (via microprocessor functions) an optimal
decoupling capacitance using one of the linear problem formulation
techniques discussed in the present invention. Thereafter, the
computer system (30) outputs the optimal decoupling capacitance
information (34) via some user-readable medium, e.g., monitor
display, network path, etc., where the optimal decoupling
capacitance includes at least a percentage of the available
capacitance area that can be used for thin-oxide capacitance
instead of thick-oxide capacitance. The computer system (30) may
additionally output the amount of leakage current resulting from
the optimal decoupling capacitance determination.
[0030] Those skilled in the art will appreciate that in other
embodiments, a software program capable of generating optimal
decoupling capacitance information consistent with the linear
optimization formulation techniques presented in the present
invention may be used. The software program may also be capable of
determining leakage current and power values corresponding to the
generated optimal decoupling capacitance information.
[0031] Advantages of the present invention may include one or more
of the following. In some embodiments, because decoupling
capacitance on an integrated circuit may be optimized using a
linear optimization formulation technique in accordance with the
present invention, integrated circuit performance may be
improved.
[0032] In some embodiments, because a linear optimization
formulation technique in accordance with the present invention may
be used to determine an optimal assignment of thin-oxide and
thick-oxide capacitance, valuable time that would otherwise be used
to determine an optimal capacitance is saved.
[0033] In some embodiments, because a linear optimization
formulation technique in accordance with the present invention may
be used to determine an optimal capacitance for an integrated
circuit or regions thereon subject to leakage power constraints of
the integrated circuit or regions thereon, capacitance may be
maximized while meeting a leakage power budget of the integrated
circuit or regions thereon.
[0034] While the invention has been described with respect to a
limited number of embodiments, those skilled in the art, having
benefit of this disclosure, will appreciate that other embodiments
can be devised which do not depart from the cope of the invention
as disclosed herein. Accordingly, the scope of the invention should
be limited only by the attached claims.
* * * * *