BlockImplementation181 Lab
BlockImplementation181 Lab
Implementation Flow
Rapid Adoption Kit (RAK)
Note: RAK Testcase database, Scripts and references can be found at ‘Attachments’ and
‘Related Solutions’ sections below the PDF.
Cadence Trademarks
Trademarks and service marks of Cadence Design Systems, Inc. (Cadence) contained in this document are attributed to Cadence with
the appropriate symbol. For queries regarding Cadence trademarks, contact the corporate legal department at the address above or call
800.862.4522.
Other Trademarks
Open SystemC, Open SystemC Initiative, OSCI, SystemC, and SystemC Initiative are trademarks or registered trademarks of Open
SystemC Initiative, Inc. in the United States and other countries and are used with permission.
All other trademarks are the property of their respective holders.
Confidentiality Notice
No part of this publication may be reproduced in whole or in part by any means (including photocopying or storage in an information
storage/retrieval system) or transmitted in any form or by any means without prior written permission from Cadence Design Systems,
Inc. (Cadence).
Information in this document is subject to change without notice and does not represent a commitment on the part of Cadence. The
information contained herein is the proprietary and confidential information of Cadence or its licensors, and is supplied subject to, and
may be used only by Cadence customers in accordance with a written agreement between Cadence and its customers. Except as may
be explicitly set forth in such agreement, Cadence does not make, and expressly disclaims, any representations or warranties as to the
completeness, accuracy or usefulness of the information contained in this document. Cadence does not warrant that use of such
information will not infringe any third party rights, nor does Cadence assume any liability for damages or costs of any kind that may
result from use of such information.
RESTRICTED RIGHTS LEGEND Use, duplication, or disclosure by the Government is subject to restrictions as set forth in
subparagraph (c)(1)(ii) of the Rights in Technical Data and Computer Software clause at DFARS 252.227-7013.
UNPUBLISHED This document contains unpublished confidential information and is not to be disclosed or used except as authorized
by written contract with Cadence. Rights reserved under the copyright laws of the United States.
The goal of this tutorial is to provide you a small example of using the Innovus software.
It is very basic by design so we highly recommend users attend one of several Innovus
training classes provided by Cadence Educational Services. For more information on
available training please visit www.cadence.com and click Services->Training.
In addition to this tutorial the following resources are recommended to continue your
Innovus learning.
Online documentation is available via https://support.cadence.com
Resources Product manuals Innovus18.1
Testcase database, Scripts and references can be found at ‘Attachments’ and ‘Related
Solutions’ sections below the PDF.
This pdf can be searched with the document 'Title:
Introduction to Innovus Implementation System 18.1 & Block Implementation Flow
on https://support.cadence.com
2. Extract the RAK database and change directory to the work directory:
4. Start Innovus:
linux% innovus
Innovus creates three files for storing commands and their output. Each of these files will
have a number at the end of the name which is incremented with each session.
• innovus.cmd – Contains list of commands executed during the session. This file
can be used to create scripts to automate the execution of the commands and learn
what text command correspond to commands executed through the GUI.
• innovus.log – Contains basic information output from the executed commands. The
commands in the file are preceded with <CMD> in the file.
• innovus.logv – Similar to innovus.log but contains more verbose amount of output.
Useful for debugging.
Tips:
• Overwrite the default names by using the –cmd and –log options when
executing innovus.
• Use innovus –init scriptName to execute a script when invoking
innovus. Type win to open the GUI when the script completes.
• Use innovus –nowin to invoke Innovus in non-GUI mode.
The Linux shell Innovus is invoked from becomes the console where standard output is
printed, and is also where you enter text commands:
Tip: When entering text commands you can utilize up/down arrows to cycle through the
command history as well as tab completion to complete commands and their options to
save on typing.
Design views
Selectability Toggles
Visibility Toggles
status
Tip: You can toggle the GUI on and off using the commands win and win off. These
commands do not work when innovus –nowin is used.
Design Views
The three design views in Innovus allow certain objects to be viewable and selectable. You
switch between views using the following widgets:
FloorplanView Widgets:
• Use the Shift key for multiple object selections, and to move multiple objects. You
can also left-click and drag the mouse to select objects, such as partition pins and
block pins.
• Use the Space Bar to change the highlighting focus on an object.
• Double-click the left mouse button on an object to view or change object attributes.
Right-click and drag the mouse to specify an area that you want to see in greater detail.
When you release the mouse button, the display zooms in to the selected area.
Click the middle button of your mouse to pan the viewable window to the center point.
This is equivalent to using the panCenter command.
Move the scroll wheel of your mouse to pan and zoom the design:
4. On the Design tab select the Binding Key button. Here you can view, edit and add
binding keys.
Tip: To open the Attribute Editor on overlapping objects, first click the LMB directly over
the object, then use the Space Bar key to cycle through the objects. Once the desired object
is selected, enter the Q key.
2. Select one of the blocks in the floorplan. Observe the instance name is shown in the
bottom left corner of the GUI.
4. At the top the Attribute Editor it shows the Object Type followed by the attributes and
their values. Notice the nets connected to the block are selected.
The Auto Query (Q) button located at the bottom of the design display area enables and
disables the auto query. When enabled, information on the object the cursor hovers over is
shown in the lower left corner of the GUI. Auto Query is useful to quickly find out object
information without having to select the object and open the Attribute Editor.
If there are overlapping objects under the cursor, text information displays for the object
that is on top. Use the N key to get information on the next object, and the P key to get
information on the previous object.
Tip: Press the F8 when hovering over an object to print the property information to the
console window. Note if a menu appears when you press F8, select Send F8.
• Cadence Help provides access to the documentation including the User Guide,
Text Command Reference, Menu Reference and LEF/DEF Language Reference.
Open Cadence Help by selecting Help - Documentation Library or executing
cdnshelp from the Linux shell. Clicking the Help button on any of the Innovus
forms will also open it. Select Help - Launch Demo from within Cadence Help
to see a demonstration of using it.
• Command line help is available from the Innovus console using the man or help
commands.
7. Use the man command to display the man page for the optDesign command:
man optDesign
Press the Spacebar or Enter keys to scroll down and Q to quit the man page.
help set*Mode
help setNanoRouteMode
Use the Design Browser to navigate through the chip's design hierarchy. You can use the
Design Browser to view the design hierarchy tree at any time after the design is imported.
The Design Browser also makes it easier to highlight specific modules, instances, or nets
in the design display area.
You can use the widgets in the Design Browser to open forms to navigate through
displays, and perform actions. From the Design Browser, you can access the Connectivity
Browser to display the number of nets between instances, and the Attribute Editor to
display an object's type, name, and attributes.
1. Make sure nothing is selected and then open the Design Browser by selecting Tools -
Design Browser. The window below appears.
2. The root of the Design Browser is the top cell of the design, specified as the Hier
Cell. Note if objects were selected, they would be listed in the Design Browser rather
than the top cell. List the Modules at the next level of hierarchy by click the plus sign
‘+’ next to Modules.
For each module it lists the instance name (i.e. proc0) and the module name in
parenthesis (i.e. proc) following by the number of leaf cells and blocks in the module.
4. Double-click on the instance proc0 in the Design Browser to drill down and list the
objects belonging to it. Notice proc0 is listed as the top module now.
The search box is used to search for instances, nets, groups or cells and accepts
wildcards.
5. Set the object type to Net and enter the net name proc0/rfo[11] followed by the
Enter key. The net matching that name is listed in the Design Browser. You can
expand it to see what pins it connects to.
You can also load selected objects into the Design Browser.
7. Fit the design and select the block in the lower right of the floorplan.
8. Click the Get Selected widget in the Design Browser and observe the selected
object(s) is listed.
widget to select all the items in the GUI that are in the list.
2-12 Stylus
The Stylus Common User Interface (Common UI from this point onwards) has been
designed to be used across Genus, Joules, Modus, Innovus, Tempus, and Voltus tools. By
providing a common interface from RTL to signoff, the Common UI enhances user
experience by making it easier to work with multiple Cadence products.
The Common UI provides an improved interface with reduced number of commands and
attribute cleanup. It also enhances usability with cleaned up log files and improved
messaging.
You can now use a single command, get_db , to query and filter all database
attributes. get_db replaces dbGet, get_ccopt_property, and various
other get methods. It includes get_property timing attributes,
although get_property is still retained for SDC usage.
The set_db and reset_db commands are the companion commands to set values.
Unified Metrics:
Unified Metrics is a system integrated with Innovus that delivers data about the design
and run to you. It displays the streamed data from the reports obtained during the flow in
a structured format.
For more information on Stylus, please refer Innovus 18.1 User Guide.
This will create a file called DBS/design.enc which contains the command to restore
the database directory design.enc.dat.
In a new session you can then restore the design using one of the following methods:
The Innovus design database is composed of several ascii files representing the different
design data. The files may vary depending on the state of the design:
exit
1. Invoke Innovus:
innovus
Use the Design Import form to import the Verilog netlist, physical libraries (LEF),
process technology libraries, timing libraries, and timing constraints.
Observe the Design Import form is populated with the settings from leon.globals.
Tips:
• When entering the values for the first time use the Save button to save the settings
to a file. Then you can load the file in future sessions.
• The technology LEF file must always be listed first in the list of LEF files.
• Use the equivalent text commands to the Design Import form to quickly load an
existing globals file:
source DATA/leon.globals
init_design
8. Press the F key to fit the design and you should see the floorplan below. The
floorplan has the blocks placed and power routing complete. You are now ready to
begin the implementation flow.
First, it is important to specify the process technology because it sets capacitance filters
and extraction effort level based on the process node.
setDesignMode –process 45
Using multiple CPUs can significantly decrease run time to implement a design. Two
CPUs can be used with the base license. Additional licenses are needed beyond two
CPUs as explained in the Accelerating the Design Process By Using Multiple-CPU
Processing chapter of the Innovus User Guide.
10. If the machine you are running on has two CPUs available select Tools – Set
Multiple CPU Usage.
12. Click OK. This will enable the use of two CPUs by commands in the flow that run
multi-threaded. You’re welcome to set this value to a higher number of CPUs if you
have the resources and licenses available.
14. In the Timing Mode select On-Chip Variation and select CPPR.
Alternatively, placement and optimization can be run separately using the commands as
follows, but place_opt_design is recommended as it provides better integration
between placement and optimization to achieve faster runtime and better PPA.
placeDesign
optDesign –preCTS
1. You can specify placement and optimization mode settings prior to running
place_opt_design. Mode settings allow you to customize how commands run.
Select Tools – Set Mode – Mode Setup to specify the mode settings for placement
and optimization.
• Select Place IO Pins if it’s not selected so that the IO pins get placed.
• Select Optimization to see the optimization settings. You will use the default
settings for optimization.
Tip: The equivalent text commands are setPlaceMode and setOptMode to specify
these mode settings. getPlaceMode and getOptMode report what the current settings
are. Other commands have similar set*Mode and get*Mode commands.
4. Specify cells with small drive strength should not be used by running the following
setDontUse commands:
enable_metrics –on
place_opt_design
Once placement and preCTS optimization is done, you can view the placed design in the
Amoeba and Physical views. Observe the placed standard cells are visible in the Physical
view (you may have to zoom in or turn off the display of Nets to see the placed standard
cells). The reason there are routed nets is because placement automatically calls Early
Global Route (eGR) so you can analyze congestion and estimate parasitics. In the next
section you’ll manually run Early Global Route to analyze congestion.
In the console you’ll see a timing summary output so you can quickly see timing results
after optimization as well as standard cell utilization (density) and EarlyGlobalRoute
overflow values:
If there are any placement violations, they should be reviewed and fixed
saveDesign DBS/prects.enc
Note: The check_design command checks the preconditions for major flow steps
before they are run. If there are errors identified by check_design, the current script
will stop check.
Option -type can be used to specify the category of the flow to be checked. For
example, check_design -type place checks for possible issues related to placement
such as PG rail alignment and pin access issues. This check is run before placement.
The red color diamond shapes which are Early Route congestion markers. Zoom-in to view
the congestion markers. Inside the diamond marker, there is a set of numbers in the format
of either H: #-top/#-bottom or V: #-top/#- bottom. H stands for horizontal congestion
and V stands for vertical congestion. The #- top is for the required number of routing tracks
used in this area and #-bottom is the available routing tracks. These diamond shaped
congestion locators represent an average in the area. Since there is only 1 vertical route
layer (metal 2), most congestion markers are vertical.
Another display of congestion will color the area based on the level of over congestion.
This makes it easier to distinguish whether an area is highly or moderately congested.
Observe a new pane appears on the left side of GUI. This pane has controls for
displaying horizontal and vertical congestion separately, as well as how many tracks a
GCell is overcongested. For example, green indicates overcongestion of 3 tracks. Zoom
and pan around to see areas of overcongestion.
8. Make nets visible again by selecting the visibility checkbox for Net on the Layer
Control bar.
9. Another way to analyze congestion is to view the log file or the Innovus console for the
congestion table produced by EarlyGlobalRoute. The label of the table is:
Congestion distribution:
For details on how Overflow values are computed, refer to section titled
“Congestion Distribution Report” in Innovus User Guide 18.1
Local HotSpot Analysis: normalized max congestion hotspot area = 92.13, normalized total
congestion hotspot area = 204.13 (area is in unit of 4 std-cell row bins)
These top hotspot scores indicates that congestion is localized in the mentioned bound box
area. Analysis must be done to avoid these congestion hotspots.
13. Click OK. Observe the overflow is much more reasonable (actual numbers will vary
but should be near 0%):
14. Run command reportCongestion –hotspot. The hotspot values are returned 0
as shown below,
[hotspot] +------------+---------------+---------------+
[hotspot] | | max hotspot | total hotspot |
[hotspot] +------------+---------------+---------------+
[hotspot] | normalized | 0.00 | 0.00 |
[hotspot] +------------+---------------+---------------+
Extraction is run in pre-route mode prior to signal routing and in post-route mode after the
signals are routed with NanoRoute. In post-route mode there are four effort levels to choose
from (low, medium, high and signoff) which increase with accuracy at the expense of
longer run-times.
The RC extraction mode can be changed by the Tools - Set Mode - Specify RC Extraction
Mode form. Since the design has not been routed we will leave the default mode set which
is pre-route mode.
2. Click OK.
If you review the log file you’ll see extraction is run and then a SPEF file is written out.
The equivalent text commands to run extraction and export SPEF are:
extractRC
rcOut –spef leon.spef –rc_corner rc_worst
When timing analysis is done, a timing summary is printed to the Innovus shell. This file
is saved under directory timingReports. The timingReports directory contains several
additional reports commonly used for debugging:
The file with suffix .summary.gz, is created after each phase once timing analysis is
run. As shown below, file leon_preCTS.summary.gz, shows timing and DRV
information at preCTS stage (In Pre-CTS stage focus is on setup whereas hold is ignored).
File leon_preCTS_default.tarpt.gz shows details of the top critical setup paths in the
design.
4. After running timing analysis you typically want to explore and debug certain timing
paths. Run report_timing to report timing for the worst path.
report_timing
5. Close the metric track for pre_cts by running the following command
pop_snapshot_stack
The Global Timing Debug Browser provides a visual display of the timing paths. This
browser is a powerful debug tool providing cross-probing between the report information
and the Innovus design display area. You can group timing violations into categories
making it easier to debug timing paths in groups rather than individual paths. Perform the
following to open the Global Timing Debug Browser:
3. Click OK.
A Timing Debug form displays path histogram and a Timing Browser displays in the
Innovus main window.
3. Double-click the first path in the Path List to open it in the Path Analyzer.
• Header shows start point/end point, slack, clock edges involved, skew and view.
• Slack Calculation shows components of slack calculation for launch and capture
paths.
• Use tabs in the center of the form to show different details of the path. Selecting
an instance or net highlights it in the GUI.
• Data delay part shows proportional delay for each instance and net in the path.
Hover over a bar to show delay value and percentage of total path delay.
• Hierarchy View shows hierarchical boundaries the path crosses.
This tutorial does not go into detail on using Global Timing Debug but the Rapid Adoption
Kit (RAK) titled Global Timing Debug (GTD) using Innovus or Tempus takes you through
a lab showing the details of debugging timing using GTD.
• Define a non-default rule (NDR) that has double the width and spacing of the
default rule. This will be used to route the clocks.
• Configure CCOpt and generate a CCOpt specification file.
• Run CCOpt-CTS to build the clock tree.
2. On the Create Non Default Rules form you can seed the width and spacing fields with
the values of an existing rule. The default width and spacing are shown by default.
Specify the rule name 2w2s and specify double the width and spacing of the default
rule as shown below:
4. Create a route type to define the NDR and layers to use for routing the clock tree:
5. Specify this route type should be used for trunk and leaf nets:
push_snapshot_stack
9. Run CCOpt-CTS:
ccopt_design –cts
10. Save the design by typing the following at the Innovus command prompt:
saveDesign DBS/cts.enc
1. Run post-CTS timing analysis by typing the following command in the Innovus
console:
timeDesign –postCTS
optDesign –postCTS
3. Run post-CTS hold timing analysis by typing the following command in the Innovus
console:
pop_snapshot_stack
saveDesign DBS/postcts.enc
The CCOpt Clock Tree Debugger is a graphical tool for analyzing and debugging the
clock tree results.
1. Select Clock - CCOpt Clock Tree Debugger. The CCOpt Clock Tree Debugger
appears.
2. The dialog box CTD Configuration is shown. Click OK without changing any
settings.
4. Display the Key Panel by selecting View -Key Panel or by selecting the Key tab in
the upper-left corner.
The Visibility menu filters which colored objects are to be made visible.
5. Enable the check box next to Visibility - Cell type -Clock sink to disable the
coloring of all the clock sinks.
7. Select an instance or net in the debugger and observe its selection in the Innovus
Implementation System GUI.
The Control Panel combines the functionality of the Visibility and Color by menus into
a single form.
8. Select View - Control Panel or click the Control tab in the upper-right to open the
Control Panel.
The radio buttons specify what the coloring is based on. You can change the color
manually.
10. Enable the Clock tree radio button. The clock tree will be colored with the default
color.
11. Enable the Timing windows check box in the Control Panel
12. Select No Color in the Control Panel in preparation for the next steps.
14. In the Browser, minimize the Analysis Views except for slow_max:setup:late, by
clicking the ‘-‘ symbol
17. Clear the highlighting by clicking the RMB in the display area and selecting
Dehighlight All.
18. The Path Browser enables you to see the detailed path delays for the min and max paths
push_snapshot_stack
Running NanoRoute with the SI Driven option requires the capacitance table file (or QRC
technology file), and this file was read in during design import. Choosing SI Driven option
in NanoRoute is our first line of defense against noise.
saveDesign DBS/route.enc
At the Innovus prompt run Post-Route timing analysis to report any setup or hold
violations. The following increases the RC extraction effort level to medium so turbo-QRC
is run which correlates better to signoff extraction. Note an effort level of medium or higher
requires a QRC technology file and effort level of high or signoff also requires a QRC
license.
3. Run setup and hold optimization simultaneously to fix any remaining timing violations:
A small number of violations may remain. If so, run optDesign again as above and
they should be resolved.
pop_snapshot_stack
saveDesign DBS/postroute.enc
7. We can generate the reports from the specified metrics by running the following
command
10. First select specific area which you want to open in 3D view using RMB. Next, on the
right-top there select the 3D view option. In the dropdown, choose Base on selected
3-D View
1. Select Verify – Verify Geometry to verify physical design rules. Verify Geometry is
used for designs above 20nm. For 20nm and below design use Verify DRC.
2. Click OK.
Review the summary in the console once Verify Geometry completes. Are there any
violations reported? Use the Violation Browser to debug the violation.
4. Select the violation in the Violation Browser as shown below to zoom to the
violation.
Tips:
• Double-clicking a violation marker in the GUI will automatically open the
Violation Browser with that violation selected.
• Select Load Violation Report to load a violation report from a physical
verification tool for debugging in Innovus.
Do the following to manually move the wire and fix the violation.
2. Click OK. This will generate a default mapping file called streamOut.map and then
export GDS based on this mapping file. You can open streamOut.map to view the
example mapping file.
3. Click OK. This will copy the libraries and design data currently in memory to a
directory called testcase then tar and gzip it into the file testcase.tar.gz. The design can
then be easily restored by extracting the file and running innovus –init
testcase.tcl.