Chapter 1 A Quick Tourfr om Gates to Layout

Chapter 1 A Quick Tourfr om Gates to Layout
Copyright © 2007 by Lee Eng Han.
All rights reserved. No portion of this book may be resold without the
written permission of the author.
For queries about permission to reproduce selections from this book,
please send an e-mail to Eng Han at [email protected]
Visit for information about this book.
Table of Contents
Version 1.0
1. A Quick Tour from Gate to Layout ............................. 1
Step 1 Load Library .................................................................. 2
Step 2 Import Gate Netlist ........................................................ 5
Step 3 Specify Design Constraint .............................................. 8
Step 4
Floor Planning .............................................................. 10
Step 5 Power Planning .............................................................14
Step 6 Physical Synthesis .........................................................16
Step 7 Clock Tree Synthesis ................................................... 19
Step 8 Routing ......................................................................... 23
Step 9 Physical Verification .....................................................26
Step 10 Post-Layout Verification ...............................................28
Summary .................................................................................... 29
Chapter 1
A Quick Tourfr om Gates to
If you have no prior knowledge of Place and Route (P&R), this chapter
is designed to bootstrap you into the design of digital layout. Later in the
book, you will be able to master the rest of the materials. This chapter
covers the essential steps of a Gate to Layout flow. The details of the
flow will be addressed in subsequent chapters. If you are experienced in
P&R, you still might want to browse through this chapter to get a feel
for the technical aspects covered in the other chapters.
The basic steps of a Gate to Layout flow include the following:
Step 1
Load library
Step 2
Import gate netlist
Step 3
Specify design constraints
Step 4
Floor planning
Step 5
Power planning
Step 6
Physical synthesis
Step 7
Clock tree synthesis
Step 8
Step 9
Physical verification
Step 10
Post-layout verification
Most P&R projects have to move through these steps. However, the
steps are not necessarily executed in the order listed. Below are
examples of the variation that may occur in the execution of P&R:
 Physical verification (step 9) of the pad ring during floor
planning (step 4).
 Physical synthesis (step 6) to assess the feasibility of a floor plan
(step 4) without power planning (step 5).
 Some P&R flow perform physical synthesis (step 6) and clock
tree synthesis (step 7) concurrently.
Not all the steps are necessary for a P&R project. Depending on the
logic design requirements and fabrication process technology, some
steps can be omitted, or some steps can be added. For example,
 Clock tree synthesis might not be necessary at 0.6um process
 Crosstalk noise violation analysis and fixing is a must for layout
using 0.18um or smaller process technology.
Step 1: Load Library
A P&R library contains two types of information
 technology library
 cell library
Step1.1 Technology Library
Wires that satisfy all layout design rules must be put in place by the
router. The timer engine requires accurate parasitic capacitances and
resistances for static timing analysis, crosstalk analysis and power
analysis. Information regarding the layout design rules and capacitance
look-up table are nested in the technology library. The following table
lists some of the P&R tasks and their corresponding roles from the
technology library that are necessary to perform the tasks.
P&R Task
Technology library
Congestion driven P&R
Process design rules.
Timing driven P&R
Routing Parasitics.
Cross-talk aware P&R
Routing parasitics with coupling
Electro-migration limit for each metal
Metal and via density requirement
Electro-migration (EM)
Metal and via density
Step 1.2 Cell Library
The cell library holds both logical and physical information of the logic
cell. The logical information is similar to the contents of a synthesis
library. Several different types of information exist and cannot all be
listed here. The following are some of the common contents of the cell
 cell type (e.g. combinational, sequential, pad, timing model, etc.)
 Pin-to-Pin delay
 slew for output pin
 capacitive loading on input pin
 leakage and dynamic power consumption
 design rule (e.g. maximum input slew for input pins and
maximum load for output pins)
 maximum allowable noise for input pins, and holding resistance
for output pins
The full physical layouts of the cells are too complicated to be used in a
P&R environment. Hence, the physical information in the cell library
contains a simplified version of the layout commonly known as an
“abstract”. An abstract contains the following information:
 cell name
 size of the cell
 allowable orientation
 pin names and their layout geometries
 routing blockages
 process antenna areas for the pins
Power and ground pins are typically excluded from the logical library
but they must be included in the abstract library. The following graphic
(Fig. 1.1) depicts an example of a standard cell abstract.
Figure 1.1 : An example of a standard cell abstract.
A popular format for abstract library is Layout Exchange Format (LEF).
LEF format supports both the technology library and the cell library.
The technology defines the name of the layers (e.g. metal layers and via
layers) that are used in the cell libraries, therefore it is mandatory to load
the technology library before the cell library.
Step 1.3 Three Types of Cell Libraries
There are three types of cell libraries:
 Standard cell library
 Pad cell library
 Macro library
Standard cells must be placed in the “core” and on the “cell row”.
“Core” and “Cell row” are described in the latter part of this chapter.
Pads and macros do not have this restriction and can be placed anywhere
on the layout. However, pads are typically placed on the peripherals of
the layout. All P&R tools have specific functions to automate the
placement of the pads.
Step 2: Import Gate Netlist
Verilog is the most popular gate-level netlist format. It is also the
preferred netlist format for most P&R tools. An alternative is the use of
the VHDL gate-level netlist.
After loading the netlist into the P&R tool, the logic gate in the design
bind to their cell master in the cell libraries. This process seems trivial,
but it is necessary to cater to all situations where cells from different
libraries have the same cell name.
Step 2.1 Cell and instance
A cell library is a collection of cells. Standard cell refers to a logic gate.
I/O1 cells are usually called I/O pads. Hard macros refer to the layout of
the IP2 . An IP without a layout implementation is called a soft macro.
An instance refers to a cell in the design. Rather than saying “adding the
cell AND2D1 to the design”, the common term used is “instantiating the
cell AND2D1 to the design”. Every instance in the same design
hierarchy must have a unique instance name. The following example is
an instantiation of a 2-input AND gate:
I/O stands for Input/Output.
Intellectual Property.
AND2D1 inst0 ( .A1(net1), .A2(net2), .Z(net3) );
The instance name of the 2-input AND gate is “inst0”, and the master
name (or cell name) is AND2D1.
Figure 1.2 summaries the definition of cell and instance.
Figure 1.2 : Cell and instance.
Step 2.2 Define power and ground connections
Refer to the instantiation of a 2-input AND gate.
AN2D1 inst0 ( .A1(net1), .A2(net2), .Z(net3) );
Gate level netlist does not typically include the connection to the power
and the ground supplies. However, these connections have to be defined
before layout implementation. All P&R tools use global nets 1 and
wildcards to define these connections. For example, the following
Design Exchange Format (DEF) statements create two new nets named
VDD and VSS. It also connects all the pins with name “vdd” to net
VDD, and all the pins with name “vss” to net VSS. The “*” in the
commands is the wildcard that matches to any instance name.
-VDD (* vdd)
-VSS (* vss)
If VDD and VSS are the only two power nets in the design, and all the
power pins of the cell instances are named “vdd” and “vss”, then these
two commands are sufficient to define all the power supply connections.
Tie-high and tie-low refer to the connection of the input pins to the
power supply. Tie-high and tie-low are represented as 1’b1 and 1’b0 in
the Verilog netlist. There are two ways to physically connect tie-high
and tie-low nets. These nets can either connect directly to the power and
the ground nets, or connect using tie-high and tie-low cells. Whichever
the case, it is the design of the pad library and the standard cell library
that determines the appropriate type of connection.
Insertion of Tie Cell into the design
Can you suggest the advantages of inserting the tie cells into the
design after placing the instances, rather than inserting them into the
design during logic synthesis?
Step 2.3 Instance name and hierarchical instance name
A design can contain many hierarchical levels. The design name is the
same as the top-level design hierarchy name. An instantiation can be a
cell or a sub-hierarchy.
The net is global because it can be reference directly by its name at any design
hierarchy without the use of the full hierarchical name.
A netlist can instantiate the same sub-hierarchy several times. However,
in order to allow the layout of the sub-hierarchy to be implemented
differently for each instance, the sub-hierarchy must be “uniquified” by
duplicating the sub-hierarchy with different hierarchy cell names. Figure
1.3 illustrates the “uniquification” process. Note that instance names are
preserved during uniquification.
Figure 1.3 : Uniquify the netlist.
A full hierarchical name is required to refer to a particular instance.
Refer again to Figure 1.3. Assume there is a flip-flop with instance name
“U1” in “Adder32”. As “Adder32” is instantiated four times before
uniquification, we can refer to that flip-flop in the four instances of
“Adder32” as Inst0/U1, Inst1/U1, Inst2/U1 and Inst3/U1. As instance
names are preserved during uniquification, the instance names of the
four flip-flops remain unchanged after uniquification.
Step 3: Specify Design Constraint
P&R is a constraint-driven process. In the absence of constraints, P&R
optimization is purely congestion-driven. A timing constraint is an
important part of the design constraint. Timing constraints specify the
timing goals of the design. In order to perform timing-driven placement
or physical synthesis, timing constraints must be available. Timing
constraints are most likely to be specified in SDC format.
The timing constraints should be specified at the top-level of the design.
It is necessary to specify a complete top-level timing constraint for a
timing-driven flow. Any unconstrained timing paths will not be
optimized for timing performance. Hence, unconstrained paths might be
implemented with logics that are very slow or with extraordinarily large
On the other hand, over-constraining the timing requirement is
undesirable. An over-constrained design can result in an implementation
that is unnecessarily large in area and the P&R will possibly consume a
much longer run-time.
In addition to timing constraints, there are constraints not related to
timing performance. There is no standardized name for this type of
constraint, so it will be termed a “non-timing constraints” in this book.
There are many different types of non-timing constraints. The following
list describes some of them.
 design rules which include maximum fan-out, maximum slew
and maximum capacitance loading
 scan-chain re-ordering and re-partitioning
 selective hierarchy flattening
 buffering of inputs and outputs with user-specified cells
 identification of cells that the tool cannot modify or can only
 identification of nets that must be preserved during logic
 disallow the use of certain cells
 assign higher priority to certain nets so as to achieve shorter
wiring length
 restriction in the area that certain cells can be placed
 among others
Non-timing constraints can be employed to ensure the physical
implementation meets the design requirements, improvement of layout
quality and turn-around time, as well as to work-around the limitations
of the P&R tools.
Step 4: Floor Planning
Floor planning is the first step of physical layout implementation. A
floor plan should include the following decisions:
 size of the layout
 core area
 placement of I/O pads and I/O pins
 placement of the hard macros
A floor plan should include the placement of all the pads (or pins) and
the hard macros. However, the standard cells are not placed yet and no
routing is performed at this stage1.
Step 4.1 Size of the layout
The first step in floor planning is to define the outline of the layout. If
the layout is rectangular, only the length and the width of the layout are
required. More co-ordinates are needed to define the outline of a
rectilinear layout, such as an L-shape layout. Most of the P&R tools do
not alter the size of the layout specified by the user.
Step 4.2 Core Area
The core area is usually defined by specifying the distance between the
edge of the layout and the core, as shown in Figure 1.4.
It is possible to pre-place some standard cells, and also to pre-route wires at this stage
of the flow. It is also possible that smaller macros are left unplaced.
Figure 1.4 : Definition of the core area.
All standard cells must be placed in the core area. I/O pads and macros
do not have this restriction although it is common to place macros in the
core area. The area outside the core can be used to place the I/O pads,
the I/O pins, and the core power rings.
Standard cells are placed in rows, similar to placing books on a book
shelf. The rows are called “cell rows” and are drawn inside the core area.
All cell rows have the same height. There are three common ways to
arrange the cell rows (Figure 1.5).
The most common approach for layout with more than three metal
layers is to flip every other cell row which does not leave a gap between
the cell rows.
The second configuration is to flip every other cell row, but leave a gap
between every two cell rows. The purpose of the gaps is to allocate more
resources for the inter-connect routings.
The last configuration is to leave a gap between every cell row, and not
flip the cell rows. This configuration is useful when only two or three
metal layers are available for routing.
Figure 1.5 : Three types of cell row configuration.
The “slanting lines” on the right of the cell rows in Figure 1.5 denote the
orientation of the cell rows. Modern P&R tools will fill the core area
with cell rows automatically. Some P&R tools require the user to
specify the areas in the core where the cell-row should be created.
Arrangement of the cell rows
What is the reason for flipping the cell rows?
When you need to leave a gap in between the cell rows, how do you
determine the height of the gap?
If all the standard cells are in the same power domain, then only one
core area is required. In a multiple core power P&R flow, more than one
core area must be defined, and every core area must associate itself with
a power domain.
Step 4.3 Placements of IO Pads and IO Pins Geometries
For a chip-level layout, the next step is to place the IO pads. The P&R
tool can fill the gaps between the pads with pad filler cells and corner
cells. For a block-level layout, the user needs to define the location and
geometries (size and metal layer) of every IO pin.
Step 4.4 Placements of the Hard Macros
The floor plan is complete if the design does not contain any hard
macro1. Otherwise, the next step is to place the hard macros. Placing the
hard macros may not be a simple task. A good macro placement has the
following qualities:
 provides a compact layout
 does not cause routing congestion
 does not make timing closure difficult
 allows robust power routing between the power pads and the
The biggest challenge in placing the macros is in assessing the quality of
the floor plan, which cannot be achieved without executing the rest of
the flow. Thus, floor planning is an iterative and time consuming
process. The trick in performing floor planning is to shorten the turnaround time of the iterations, and to reduce the number of iterations. The
following figure (Figure 1.6) depicts a simple floor plan of a chip-level
layout with only one macro.
Examples of hard macro are memory, analog layout or a complete layout of a CPU.
Basically, any instant that is not a standard cell is a hard macro!
Figure 1.6 : A floor plan with one macro and I/O pads
Step 5: Power Planning
All connections to the power and ground nets are routed during power
planning. The only exception is the tie-high and the tie-low nets. Most
P&R tools use a dedicated router to route the power nets. All power
routings created by the power router are considered pre-routes, and are
not modified by the detailed router when the signal nets are routed.
The key consideration for power planning is:
 an acceptable IR-drop from the power pads to all power pins
 meeting electro-migration requirements
 does not result in routing congestion
 compact layout
A power plan consists of several types of power structure. Figure 1.7
illustrates a typical sequence to construct the power structures.
1. core power rings are routed first
2. core power pads are connected to the core power rings
3. the power rings are added around the macros where necessary
4. vertical stripes and horizontal stripes are added to reduce the IRdrop at the power rails of the standard cells and the macros
5. the power pins of the hard macros are tapped to the core rings or
the power stripes
6. if tie-high and tie-low cells are not used, the tie-high and tie-low
inputs to the hard macros and IO pads are tapped to the power
7. the power rails for the standard cell are added to the power plan
The power rails can tap the power from the core power rings, the power
stripes and the macro power rings.
Figure 1.7 : Steps in building a power plan.
Priorities of the power structures
In the 7 steps illustrated above to build the power structures, can you
suggest a reason why “step 2” should not be executed last?
If you are to use both vertical and horizontal stripes, what are the
considerations to decide which one should be added to the power
plan first?
Step 6: Physical Synthesis
Physical synthesis refers to the placement of the standard cells and the
optimization of the layout base on the design constraints.
Physical synthesis consists of a few typical phases:
 global placement
 global routing
 physical optimization
 detailed placement
 further physical optimization
After physical synthesis, all standard cells are placed on the cell rows.
The placement should be legalized, which means that the standard cells
are on the cell row, on the placement grid, non-overlapping, and the
power pins of the standard cells are properly connected. The placement
should be routable, meeting the timing goal, and satisfy the placement
constraints specified by the user.
In order to meet the timing goal, the tool might need to optimize the
netlist. Different tools have different capabilities given the type of
optimization it can perform. The user can also configure the type of
optimization the tool can utilize. Some of optimization techniques a
P&R tool can employ are listed below. The optimization techniques are
listed in an increasing order of structural change relative to the original
 gate sizing
 buffer insertion and removal
 pin swapping
 cloning
 logic restructuring
 architecture retargeting
Physical synthesis becomes essential when the IC industry started to
adopt process technologies that are 0.25um and smaller. The following
table summaries the evolution of the placement methodology.
Process Technology
Physical Optimization Techniques
0.6um and larger
Placement is congestion driven. Manual
insertion of buffers to long nets after routing.
Placement is timing driven. Physical
optimization is restricted to gate sizing and
buffer insertion.
0.25um and smaller
Physical synthesis is fully adopted.
Table 1.1: Evolution of placement methodology with process technology.
Most P&R flows will not attempt to restructure the logic in the clock
network. Some P&R tools have the ability to size the cells in the clock
network during physical synthesis.
Mixing clock and data
A clock network can contain combinatorial logic cells. Clock signals
can also be used as data signals. The first example below is a glitchless multiplexer where the output clock is one of the two input
clocks. This example illustrates that there can be combinatorial gates
in the clock network.
The second example shows multiplexing of a clock signal with data
signals. The third example shows the uses of the clock to select the
output of a multiplexer.
As physical synthesis preserves the clock network, the combinatorial
gates in the clock network may not be sized or buffered
appropriately. Thus, a huge negative slack might be expected after
physical synthesis if no additional steps are taken to avoid these
To achieve a good estimation of the inter-connect parasitics, global
routing is performed during physical synthesis. It is assumed that
detailed routing will match global routing closely so that physical
synthesis is optimizing on the real critical paths.
The routing congestion map can be derived from global routing. Any
routing congestion at this stage should be resolved as much as possible
by reiterating the placement with additional controls, or by improving
the floor plan and power plan.
Before proceeding further to the layout design, the layout is now ready
to perform IR-drop analysis. Two main objectives of IR-drop analysis
are to ensure
 all power pins of the hard macros and the standard cells are
connected to the power structures
 the voltage drops in the power structures are within acceptable
Step 7: Clock Tree Synthesis
After all the standard cells are placed, the clock nets are buffered. The
following list provides the additional requirements for synthesizing
clock trees when compared to the buffering of the high fan-out nonclock nets:
 clock latency
 clock skew
 restriction on the type of buffer and inverter the clock tree can
 stricter signal slew requirements on the clock nets
Clock latency is the delay of the clock signal from the clock source to
the clock pin. Clock skew is the difference between the clock latencies
and the two clock pins.
It is straight forward to specify the clock tree requirements to the P&R
tool. If the clock tree starts from one source and fans-out only to the
clock pins of the flip-flops and the macros, the clock tree meets the
requirements of the P&R tool. Unfortunately, this is not always the case.
For example, the clock tree schematic shown below has the following
additional requirements:
 the clock latencies of flip-flops div_reg* do not have to be
balanced with the clock latencies of the other flip-flops
 a small clock skews between the flip-flops div_reg*
 a small clock skews for the rest of the flip-flops in both
functional mode and test mode
 optimize for shorter clock latencies during functional mode
Figure 1.8 : Clock tree schematics example
Strategy for clock tree synthesis
The strategy to synthesize the clock tree is highly dependent on the
features of the P&R tool you are using.
As a P&R engineers, how do you find out all the requirements of the
clock tree?
It is not uncommon for the clock tree synthesis algorithm to generate a
poor clock tree when complicated blockages exist in the layout. The
following is an example of a poor clock tree:
Poor Clock Tree
Here is an example where the algorithm for the clock tree synthesis
cannot efficiently bring the clock signal from the clock source to the
clock pins. This is due to the presence of many large blockages. The
clock latency is unnecessarily long. Poor slews in the clock nets
resulted in a large clock skew.
Can you suggest any approaches to alleviate this problem?
Additional routing requirements are often applied on the clock nets. In
order to reduce noise coupling, the route spacing to the clock net can be
doubled. Shielding the clock net is another option to reduce noise
coupling. For a clock with very high clock frequency, it may be
necessary to use multiple-via on the clock routing to meet the electromigration rules.
Before clock trees are inserted, the tool uses ideal clock latencies and
ideal clock skews for timing analysis. After the clock trees are
synthesized, the tool should use the actual clock latencies and clock
skews for the timing analysis.
Ideal Clock Latency
The default ideal clock latency is 0. How about the ideal clock
latency of a generated clock? The divided clock “CLOCK” in the
schematic below has the following specification:
create_generated_clock –name CLOCK \
–divide_by 2 \
-source [get_pins clk_div_reg/CK] \
[get_pins clk_div_reg/Q]
What should be the clock latency of “CLOCK”? Also, what is the
frequency of the “CLOCK”; should it be half the frequency of
“CLOCK_A” or halve the frequency of “CLOCK_B”?
It is now possible to analyze and fix hold violations using computed
clock latencies associated with every clock end-point after clock tree
synthesis. Hold violations should be fixed first in “best corner”
operating condition, and then in the “worst corner” operating condition.
Current P&R tools fix hold violations by adding delay to the data path.
The tool will not attempt to make changes to the clock path. It is
advisable to analyze the buffers added by the tool for hold fixing. If the
result is not satisfactory, clock trees might need to be re-synthesized
using different approaches.
Timing nightmare after clock tree synthesis!
Imagine this scenario. The timing requirement of a design is met
after the physical synthesis step. Clock tree synthesis is then
performed and all the clock trees meet the skew and latency
specifications. However, static timing analysis (STA) shows that
there are many timing paths with very poor timing slack.
Can you suggest a few reasons for the poor timing slack?
Step 8: Routing
The process of routing the non-power signals is called “detailed routing”.
After clock tree synthesis, empty space still exists in the cell rows.
These empty spaces in between the standard cells will eventually be
filled by filler cells. Fillers can be inserted either before or after detailed
routing. If the fillers contain metal routing other than the power rail,
then the fillers should be inserted before routing. Otherwise, it is best to
add the fillers after routing. Figure 1.9 shows the results before and after
filler cell insertion.
Filler cells for a standard cell library
Filler cells come in different widths. The width of the smallest filler
cell is the size of the placement grid. The widths of all the standard
cells are multiples of the placement grid size.
Do you know the purposes of the filler cells?
The filler cells usually have widths that are given as 1x, 2x, 4x, 8x,
16x, 64x, etc. The next bigger filler cell always has its width
doubled. Do you know why?
Why is it better to insert the filler cell after detailed routing?
Figure 1.9: Inserting the filler cells
Routing is performed throughout P&R flow. Powers are routed during
power planning by a dedicated router. During physical synthesis, the
signal nets are “globally routed” using the global router. The routings
are called global routes. Global routes allow the P&R tool to resolve
routing congestion and estimate the routing parasitics. After the clock
trees are synthesized, the clock trunks are routed with actual metal
geometries. This occurs before the clock trees are re-optimized. This is
then followed by detailed routing.
Detailed routing is carried out in stages. The detailed routing stages are
different in various P&R tools, but the methodology used for detailed
routing is similar. The stages are outlined below.
1. Track routing: Global routing uses a very coarse routing grid.
Track routing assigns the global routes to the actual routing
2. Detailed routing with only metal one: Connections between cells
that are placed side-by-side are possible candidates.
3. Connecting the rest of the signal nets by following the result of
track routing: The aim is to connect all the routings so that there
are none “Open”. The routings can be full of routing violations
(e.g. short) and design rule violations (e.g. metal to metal
spacing violations).
4. Resolve routing violations iteratively: The detailed router divides
the layout into regions and works within each region to clean up
the routing violations. This process iterates with a region of
different sizes and aspect ratios. The iterative process continues
until there are no more routing violations or the limit on the
number of iterations has been reached.
5. Iterating between fixing antenna violations and cleaning up
routing violations: New routing violations can be introduced
during the process of fixing the antenna violations.
6. Optimizing the detailed routes: The types of optimization to be
performed will depend on the user’s specifications and the tool’s
capability. The optimization can include minimizing wire jog
and switching of routing layers, or even out the spacing between
the routes, and replacing single via with redundant via.
All P&R tools have the functionality to perform cell-level DRC and
LVS on a routed design. However, it is mandatory to perform a fulllayer (or sometimes called full-GDS) physical verification.
How about the flows for cross-talk reliability,
electro-migration, power sign-off and yield
These topics are beyond the scope of this chapter. The methodology
to address each of these concepts is rooted in the complete P&R
flow. They will be discussed in the later chapters.
Step 9: Physical Verification
After the layout is routed, a GDS containing the design layout can be
generated from the P&R tool. Ideally, the GDS is ready to be sent for
mask making and fabrication. However, any mistakes made during the
P&R flow will be left undetected. Physical verification is an
independent process of verifying the integrity of the layout.
There are three types of physical verification:
 Design Rule Check (DRC)
 Layout Versus Schematic (LVS)
 Electrical Rule Check (ERC)
DRC checks the layout geometries for the manufacturing process. A
full DRC deck contains hundreds of DRC rules. The following are some
common DRC rules:
 spacing between geometries
 minimum and maximum width of the geometries
 density of the metals, the poly and the diffusion
 antenna violation
 via reliability
LVS checks the layout for correct connectivity between the devices in
the circuit. A circuit device can be a transistor, a resistor, a capacitor, a
diode, among others. During LVS verification, circuit devices and the
inter-connections are extracted from the layout and saved as a layout
netlist. This typically exists in a format similar to that of the spice
format. The layout netlist is compared with the post-layout design netlist
(usually in Verilog format).
ERC identifies errors in layout that are related to electrical connections.
Examples are latch-up protection, floating substrate, floating well, and
bad device connections. Currently, the ERC rules are typically
embedded in the DRC and LVS rules.
Why is the cell level DRC and LVS by the
P&R tool insufficient?
Cell level DRC and LVS are performed by the P&R tool. Can you
suggest some reasons why physical verification can detect DRC and
LVS violations that are not detected by the P&R tool?
At 0.35um or larger technology, it is possible for the layout
implementation to meet timing performance without physical synthesis
and clock tree synthesis. In this case, the pre-layout netlist is the same as
the post-layout netlist. However, P&R with 0.35um or smaller
technology definitely requires physical synthesis and clock tree
synthesis for timing closure. Hence, the P&R tool has to make
modifications to the netlist and output a post-layout netlist. As LVS
performs the comparison between the post-layout netlist and the layout,
it does not verify that the post-layout netlist is functionally equivalent to
the pre-layout netlist. Equivalent checker tools are needed to guarantee
the post-layout netlist has the same functionality as the pre-layout netlist.
The complete physical verification flow is shown in Figure 1.10.
Figure 1.10 : Physical verification flow.
Step 10: Post-Layout Verification
Sign-off tools are tools that are used to perform the final check on the
design. Due to run-time and memory usage considerations, P&R tools
have to use simplified models during parasitics extraction, static timing
analysis and other types of analyses. Hence, there is a need for sign-off
On average, the parasitics extraction result from a P&R tool should be
within a few percentage points from a sign-off parasitics extraction tool.
However, some nets can exhibit a large difference, in excess of 100%.
Similarly, the static timing engine in the P&R tool and the sign-off static
timing tool might show differences with respect to design constraints,
delay calculation or tracing of the timing path.
The sign-off tools may find violations the P&R tool misses. It may be
necessary to correlate the interpretation of the design constraints by the
P&R tool and the sign-off tools early in the P&R project. It is common
practice to over-constrain the design by a small margin so that
differences among the tools do not pose any concerns.
The result of parasitics extraction is usually stored in SPEF or SPDF
format. The extracted parasitics can be back-annotated into the sign-off
static timing analyzer (STA) tool. STA performs delay calculations and
verifies the timing performance. The STA tool can generate a standard
delay format (SDF) file that is used by a logic simulator for post-layout
sign-off simulation. The post-layout verification flow is shown in Figure
Figure 1.11: Post-layout verification flow
If the timing goal is not achieved, then a possible attempt to close the
timing is initiated. A correction to the layout by manual gate sizing and
buffer insertion is standard practice to achieve the timing goal.
Subsequently, the post-layout verification is repeated. Manual
modification on the layout is too tedious with many timing violations.
Therefore, it is also possible to back-annotate SPEF or SPF into the
P&R tool, and let the tool optimize the layout to meet the timing
requirement. Care should be taken to ensure that the P&R tool does not
create new timing violations while fixing existing timing violations.
STA and logic simulation.
Can you suggest possible reasons for the following conditions?
 STA passes but the simulation fails on the same logic path
 Simulation passes but STA fails on the same logic path
This chapter provides a brief overview of a Gate-to-GDS flow. The flow
consists of 10 steps. A technology library, cell libraries and gate netlist
are read into the P&R tool. Design constraints are then imported into the
P&R tool. The first step in physical layout implementation starts with
floor planning. This is followed by power planning, physical synthesis,
clock tree synthesis and detailed routing. After detailed routing, the
implementation of the layout is completed. Before the layout is tapedout, it has to be verified by the sign-off flow, which includes physical
verification and post-layout verification.
This chapter outlines the tasks performed in a P&R project. It does not
cover the techniques and the details of each task. In the following
chapters, we will study the specifics of these tasks.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF