Expanding the Synopsys PrimeTime

Solution with Power Analysis

Gordon Yip, Product Marketing Manager, Synopsys, Inc.

June 2006

2006 Synopsys, Inc.

Introduction
Design closure in todays advanced designs requires a delicate balance of many complex issues. Timing remains critical, but power has become important toward achieving design success. Today, power management
is a mainstream design challenge and a key concern for chip designers. Power consumption is a critical
design delimiter. It affects packaging decisions, cooling requirements, battery life, design performance, and
chip reliability. More than ever, accurate power analysis is required for designs. Designers today must consider the impact of their design techniques on power in conjunction with timing and area. A failure to analyze
power can lead to chip failure.
Power, timing, and signal integrity (SI) effects are all interdependent at 90-nanometers (nm) and below. To
achieve the highest accuracy power analysis, an accurate timing engine is required to perform accurate timing and slew calculations. Since timing parameters affect power dissipation, designers require a solution that
takes advantage of these interdependencies. In this white paper, we propose a methodology that combines
static timing analysis, signal integrity analysis, and power analysis within a single, unified environment.

Components of Power Dissipation

For CMOS designs, the total power consumed by a circuit falls into 2 major categories: static power, which
is the power consumed when the transistors are not switching, and dynamic power, which represents the
power consumed by switching logic states.

Static Power
Static power is dissipated in several ways. Some are due to the reverse-biased diode leakage from the
diffusion layers and the substrate, but the largest percentage of static power results from source-to-drain
sub-threshold leakage current. This is caused by reduced threshold voltages which prevent the gate from
completely turning off and hence allow this leakage current (Ileak).
The leakage power is dependent on the voltage, temperature and state of the transistors.
Leakage Power = V * Ileak

Dynamic Power
Dynamic power is dissipated any time the voltage on a net changes due to some stimulus. This voltage
change charges or discharges the capacitive load of the external net. Also, the voltage change results in a
short-circuit current between the N and P transistors internal to the gate. For ASIC designs, the dynamic
power consumed by the charging or discharging of the output load external to the cell is classified as
switching power, while the dynamic power dissipated within the cell is generally classified as the internal
power.
Switching Power
The switching power is determined by the capacitive load and the frequency of the logic transitions on a cell
output.
Switching Power = * Cload * V2 * f
where the total load capacitance (Cload) is the sum of the net and gate capacitances on the driving output,
and the frequency (f) is the rate of state transitions.
Internal Power
The internal power is caused by the charging of internal loads as well as by the short-circuit current between
the N and P transistors of a gate when both are on.
Internal Power = ( * Cint * V2 *f) + (V * Isc)

2006 Synopsys, Inc.

As the input signals transition, both N and P type transistors can be on simultaneously. During this time,
current Isc flows from Vdd to Gnd causing the dissipation of short-circuit power. The short-circuit power is affected by the dimensions of the transistors, the load capacitance on the output, and the transition time of the
input signals. Circuits with slow transition times can dissipate excessive short circuit power as both N and P
transistors are on for an extended period.
ASIC or Library vendors provide power models for the internal power consumption of CMOS cells, which are
characterized with different driver output loads and input signal transition times.
The diagram below describes the power components for a simple buffer cell.

Vdd

Vd d

In b
In

I sc

Ou t

I sc
I i n tsw
N

I sc

I l ea k

I sw
N

C i nt

I l ea k

C l oad

Short circuit current

I intsw Internal switching current

I leak Leakage current

G nd

G nd

I sw Switching current

Figure 1: Components of Power Dissipation.

The leakage current Ileak can vary based on the transistor states. For example, when the input signal In is
high, and the N transistor is on, the leakage will differ compared to when the N transistor is off. When a
rising signal is applied at the input, internal power is dissipated due to Isc and Iintsw. During the transition from
low to high, the N transistor turns on and the P type transistor turns off resulting in Isc from Vdd to Gnd.
Additionally, internal switching power is incurred in charging and discharging of Cint. The switching power on
the Out net is due to Isw charging and discharging Cload.

Power Analysis Requirements

In order to analyze the power consumption of a design, it is important to consider all the factors which contribute to both the static and dynamic power. These are shown in the diagram below:
In order to analyze the power consumption, the following are required:
Netlist Data
The design netlist is required to determine the design connectivity and type of cells used in the design
and to accurately compute the capacitance on the drivers.
Cell Library Power Models
In order to compute the internal power of the CMOS cells used in the design, ASIC or Library vendors
must provide cell models which specify both the static and dynamic power consumption internal to
the cell.
Signal Activity
The signal activity of a design affects both the static and dynamic power consumption. The static power
(cell leakage) is often state dependent, and the dynamic power is directly proportional to the toggle rate
of the pins.
2006 Synopsys, Inc.


Netlist
Data

Signal
Activity

Power
Models

Net
Parasitics

Power Analysis

Figure 2: Power Analysis Requirements.

Net Parasitics / Transition Times

Net parasitics (or capacitances) affect the dynamic power of the design. Switching power is directly proportional to the net capacitance. Internal power depends on both the input signal transition times, which
are determined by the net parasitics, as well as the output load, which is a combination of the net parasitics and input pin capacitances of the fanout. The example below illustrates the relationship between the
transition times and the dynamic power consumption.

Input Transition Time

In1

Large Transition Time

Net 1

U1

D1

Out

C1
DX
DX
DX
DX

Figure 3: Transition Times Affect Dynamic Power Consumption.

The internal power consumption of U1 is determined by the transition time on In1 and the capacitive load
on the output Net1, which is the sum of the input pin capacitances of the gates it drives (gate fanout) and
the net capacitance, C1. Now if the net capacitance C1 is large, the switching power of Net1 certainly will
increase, but additionally, the transition time on Net1 will be affected.
A large transition time on a net with a large number of fanout loads results in excessive internal power
consumption for all of the gates to which it is connected. This is because the internal power consumption of
the fanout gates, the registers D1-DX in this example, is considerable because the long transition time on
the net is allowing for a longer period of time in which short circuit power is being dissipated.
In summary, the accuracy of the power analysis is dependent upon the accuracy of the inputs provided. The
netlist, cell library power model, signal activity, and parasitics/transition times need to be accurate to provide
accurate power consumption analysis.

2006 Synopsys, Inc.

PrimeTime PX Power Analysis

Integrating Timing, Signal Integrity (SI) and Power Analysis

Synopsys PrimeTime PX, the power analysis extension to PrimeTime, enables users to perform full-chip,
concurrent timing, signal integrity, and power analysis in a single, easy-to-use environment. PrimeTime PX
can perform highly accurate power analysis because it can take advantage of the timing and signal integrity
engines of the PrimeTime and PrimeTime SI solutions for processing the design data, parasitic backannotation, and calculation of transition times and timing windows. Also, the power analysis engine supports
composite current source (CCS) power models, which accurately model the CMOS cell library. Furthermore,
it also supports multiple signal activity formats. Altogether, PrimeTime PX provides the best platform to
accurately use and analyze all of the key inputs required for accurate power analysis. The PrimeTime PX
solution is described below.
Netlist Data
Clearly the design netlist data provided for analysis must accurately represent the design. Within the PrimeTime PX solution, a variety of netlist formats are supported. For both power and timing analysis, the same
netlist reader is able to process netlist and store the analysis results, thereby enabling concurrent analysis
on the same design netlist data.
Parasitics/Timing
In the PrimeTime solution, parasitics (resistance and capacitance) for the nets in the design can be annotated onto the design netlist. These parasitics can either be reduced parasitics (lumped RC or reduced pi
model) or detailed parasitics (distributed RC). By annotating parasitics, especially detailed parasitics, very
accurate delay and transition times can be calculated. These accurate delays and transition times are then
used for more accurate dynamic power calculations.
Furthermore, the effect of cross-coupling capacitances on delays and slews can also be accounted for by
annotating cross-coupled parasitics and using the signal integrity analysis engine in the PrimeTime SI solution. If the PrimeTime SI solution is not enabled, then any cross-coupling is split to ground, and the crosscoupling effect will not be accurately computed.
The PrimeTime solutions integrated static timing analysis and RC delay calculation engine accurately calculates the delay and transition time for every cell, based on the input slew on the driving cell, the annotated
RC network, and the capacitive load on the fanout pins.
CCS Cell Library Models
With the advent of smaller nanometer technologies, the CCS approach of modeling cell behavior has been
developed to address the effects of very deep nanometer processes. With the use of the CCS cell library
models, delay calculation accuracy and dynamic power analysis accuracy is further improved.
For example, the CCS timing receiver model uses 2 capacitance values rather than a single lumped capacitance for the input gate capacitance, or load, of a cell. As the input waveform switches, the first capacitance
value is used for the gate load up to the delay threshold. When the input waveform reaches this threshold,
the load is dynamically adjusted to the second capacitance value. The result is this timing receiver model
provides a more accurate model of the loading effects in the presence of the Miller effect.
The CCS modeling format supports both timing and power, thereby providing a high degree of accuracy.
Signal Activity
Accurate power analysis depends on accurate signal activity. PrimeTime PX power analysis supports both
average and peak power analysis based on the signal activity provided. For average power analysis, PrimeTime PX supports the propagation of switching activity based on either tool defaults, user-defined switching
activity, or switching activity derived from a logic simulation (either RTL or gate-level), typically saved as a
SAIF (Switching Activity Interchange Format) file. For peak power analysis, PrimeTime PX requires a timing
logic simulation and generation of a VCD that captures the activity and time of every event on each net.

2006 Synopsys, Inc.

VCD Flow

Power
Models

Netlist
& Parasitics

PrimeTime PX

Peak Power
Report and
Waveforms

Vector-free Flow

SAIF Flow
Signal
Activity
(VCD)

Power
Models

Netlist
& Parasitics

PrimeTime PX

Average Power
Report and
Waveforms

Signal
Activity
(SAIF)

Power
Models

Netlist
& Parasitics

PrimeTime PX

Average Power
Report and
Waveforms

Figure 4: Power Analysis Flows.

The VCD-based analysis is extremely accurate since all the factors contributing to power consumption are
supported in an accurate form. Peak and average power can be calculated, and detailed, time-based
waveforms can be generated.
If event-based simulation activity is not available, accurate average power can still be calculated by providing
gate-level toggle rates, typically with a SAIF file. In the SAIF file, the activity information includes the number
of times the signals toggled per net, as well as the percentage of time that the signals were at a given state.
This toggle rate information enables both dynamic and leakage power to be accurately determined.
If gate-level logic simulation data is unavailable, the PrimeTime PX solution also supports RTL VCD
(generated from a RTL logic simulation), RTL SAIF (also generated from a RTL simulation), user-defined
switching, and default analysis. In this case, the activity of unannotated nodes is obtained via an internal
zero-delay simulation to compute the toggling of the outputs. With the switching activity for every node in
the design now determined, power consumption can be calculated.
Other Benefits of Integrating Timing, SI, and Power Analysis
By combining timing, signal integrity and power analysis into a single tool and environment, productivity
and time to results are improved over separate, standalone timing and power analysis tools. With a single
environment, identical operations are not repeated. For example, timing and slew calculations are not
repeated. Netlist, parasitic and constraint file reads are not repeated. And tool setup steps are not repeated.
As a result, PrimeTime PX delivers up to 2X faster time to results over separate, standalone solutions.

2006 Synopsys, Inc.

Furthermore, as an extension of the PrimeTime environment, PrimeTime PX is easy to use and adopt. Power
analysis can be performed using the familiar PrimeTime commands, user interface, reports, attributes, and
multiple debugging features.

Summary
With the transition to finer process geometries, signal integrity effects and power consumption have become
a top concern. When ignored, circuits can either fail in silicon, or not meet performance specifications. Given
the effects of parasitics and SI on timing, and the effects of timing on power consumption, designers need
an environment in which the interdependencies between power and timing analysis are accounted for.
Use of the PrimeTime environment for processing of the netlist, parasitic and timing data contributes to a
high level of accuracy when performing power analysis. The other critical factor is signal activity. If eventbased activity is available, both peak and average power can be determined. For purely average power
analysis, the toggle rate data can produce accurate leakage and average power results. If gate-level simulation activity data is unavailable, the support for RTL level switching or default switching activity propagation
provides a good estimate of the dynamic power and state-dependent leakage.

700 East Middlefield Road, Mountain View, CA 94043 T 650 584 5000 www.synopsys.com
Synopsys, the Synopsys logo, and PrimeTime are registered trademarks. All other trademarks or registered trademarks
mentioned in this release are the intellectual property of their respective owners and should
be treated as such. All rights reserved. Printed in the U.S.A.
2006 Synopsys, Inc. 6/06.CE.WO.06-14513