luke
/
magic
mirror of https://github.com/RTimothyEdwards/magic.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
magic/doc/html/extresist.html

196 lines
9.0 KiB
HTML
Raw Normal View History

Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<HTML>
<HEAD>
<STYLE type="text/css">
H1 {color: black }
H2 {color: maroon }
H3 {color: #007090 }
A.head:link {color: #0060a0 }
A.head:visited {color: #3040c0 }
A.head:active {color: white }
A.head:hover {color: yellow }
A.red:link {color: red }
A.red:visited {color: maroon }
A.red:active {color: yellow }
</STYLE>
</HEAD>
Updated the HTML command reference files to refer to magic version 8.3, since all of the title images still had 8.2, and the title text still said 7.3 or 7.4, having not been updated for a long time.
2024-01-09 22:37:00 +01:00
<TITLE>Magic-8.3 Command Reference</TITLE>
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<BODY BACKGROUND=graphics/blpaper.gif>
Updated the HTML command reference files to refer to magic version 8.3, since all of the title images still had 8.2, and the title text still said 7.3 or 7.4, having not been updated for a long time.
2024-01-09 22:37:00 +01:00
<H1> <IMG SRC=graphics/magic_title8_3.png ALT="Magic VLSI Layout Tool Version 8.3">
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<IMG SRC=graphics/magic_OGL_sm.gif ALIGN="top" ALT="*"> </H1>
<H2>extresist</H2>
<HR>
Patch the extraction <TT>.ext</TT> files with detailed route resistance
information.
<HR>
<H3>Usage:</H3>
<BLOCKQUOTE>
<B>extresist</B> <I>option</I> <BR><BR>
<BLOCKQUOTE>
where <I>option</I> may be one of the following:
<DL>
<DT> <B>all</B>
<DD> Extract all the nets.
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
<DT> <B>threshold</B> [<I>value</I>]
<DD> With no value given, returns the current lumped resistance
threshold used to determine if a network will or will not be
analyzed for resistance extraction, in milliohms. The default
<B>threshold</B> value is 10000 milliohms (10 ohms). If
<I>value</I> is given, then set the lumped resistance threshold
to <I>value</I> milliohms.
<DT> <B>mindelay</B> [<I>value</I>]
<DD> With no value given, returns the current delay time threshold
used to determine if a network will or will not be analyzed for
resistance extraction, in picoseconds. The default
<B>mindelay</B> value is zero, indicating that delay time is
not used for determining extraction. If <I>value</I> is given,
then set the delay threshold to <I>value</I> picoseconds.
<DT> <B>minres</B> [<I>value</I>]
<DD> With no value given, returns the current absolute resistance
threshold used to prune small resistances from a network when
simplifying, in milliohms. The default <B>minres</B> value
is 1000 milliohms (1 ohm). If <I>value</I> is given, then set
the absolute resistance threshold to <I>value</I> milliohms.
Note that resistances smaller than <I>value</I> may still
appear in the output netlist if the algorithm is unable to
simplify the network around the resistor.
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<DT> <B>simplify</B> [<B>on</B>|<B>off</B>]
<DD> Turn on/off simplification of resistor nets.
<DT> <B>extout</B> [<B>on</B>|<B>off</B>]
<DD> Turn on/off writing of the <TT>.res.ext</TT> file.
<DT> <B>lumped</B> [<B>on</B>|<B>off</B>]
<DD> Turn on/off writing of updated lumped resistances.
<DT> <B>silent</B> [<B>on</B>|<B>off</B>]
<DD> Turn off/on printing of net statistics.
<DT> <B>skip</B> <I>mask</I>
<DD> Don't extract types indicated in the comma-separated list <I>mask</I>
Updated all of the documentation by copying back from the website. The website documentation now points to these contents in a clone of the repository on opencircuitdesign.com, so all future edits of the command-line documentation will be made directly to the git repository. Also: Changed the precision of box values printed in microns from 2 to 3 digits after the decimal place, so that 5 nanometer grids do not get values clipped in the output.
2022-01-21 16:44:13 +01:00
<DT> <B>ignore</B> [<I>netname</I>|<B>none</B>]
<DD> Don't extract the net named <I>netname</I>. The list of ignored
nets is global, cumulative, and persistent; it will only be cleared
when the <B>extresist ignore none</B> command is issued
(option available from version 8.3.207).
<DT> <B>include</B> [<I>netname</I>|<B>all</B>]
<DD> Extract the net named <I>netname</I>. When this option is called
at least once, the behavior of the extraction changes to extract
only the nets that have been specified by <B>extresist include</B>
commands. The list of included nets is global, cumulative, and
persistent; it will only be cleared when the <B>extresist include
all</B> command is issued (option available from version 8.3.213).
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<DT> <B>box</B> <I>type</I>
<DD> Extract the signal under the cursor box on layer <I>type</I>
<DT> <B>cell</B> <I>cellname</I>
<DD> Extract the network for the cell named <I>cellname</I>
<DT> <B>geometry</B>
<DD> Extract network geometry and present as a collection of line
elements on the layout.
<DT> <B>fasthenry</B> [<I>freq</I>]
<DD> Extract subcircuit network geometry into a <B>fasthenry</B>-format
<TT>.fh</TT> file. If <I>freq</I> is specified, the file will
be customized for <B>fasthenry</B> analysis at the indicated
frequency (in Hz).
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
<DT> <B>tolerance</B> <I>value</I>
<DD> <I>Deprecated!</I> This option is no longer used by extresist.
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
<DT> <B>help</B>
<DD> Print help information
</DL>
</BLOCKQUOTE>
</BLOCKQUOTE>
<H3>Summary:</H3>
<BLOCKQUOTE>
The normal flow through layout extraction into a simulation file
treats routes as nonphysical entities, that is, with infinitesimal
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
impedence through the wires. Standard extraction generates "lumped
resistance" values, a single resistance per network node that, along
with the node capacitance
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
to substrate, provides an <I>RC</I> time constant to approximately
model the delay from point to point in the network node. The
lumped resistance model is inappropriate for analog (i.e., SPICE)
simulation, and for digital simulation, is a poor approximation
for branching networks, where the delay between endpoints is
different for each pair of endpoints in the network node. <P>
The <B>extresist</B> command provides a method for generating
a more detailed model of resistance, in which long network
routes and branching routes are replaced with resistor devices
Updated some of the HTML command reference documentation to reflect changes just made in the "extract" and "extresist" commands.
2026-01-28 23:05:00 +01:00
and device networks. This is known as "full R-C extraction". <P>
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
Updated some of the HTML command reference documentation to reflect changes just made in the "extract" and "extresist" commands.
2026-01-28 23:05:00 +01:00
Using <B>extresist</B> as a standalone command is a multi-step
process. It is first necessary to run <B>extract</B> to get
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
the initial netlist. After a <TT>.ext</TT> file has been generated,
the <B>extresist</B> command may be run. The output is a file
<TT>.res.ext</TT> for each cell in the hierarchy.
Updated some of the HTML command reference documentation to reflect changes just made in the "extract" and "extresist" commands.
2026-01-28 23:05:00 +01:00
Finally, with the option <B>extresist on</B> set, <B>ext2spice</B>
will generate the final, detailed simulation file. <P>
Prior to magic version 8.3.597, it was also necessary to run
<B>ext2sim labels on</B> and <B>ext2sim</B> to write out a
<TT>.sim</TT> and a <TT>.node</TT> file before running
<B>extresist</B>. This is no longer necessary. There is since
magic version 8.3.597 an option <B>extract do resistance</B>
that runs the resistance extraction in sequence with the regular
extraction, producing both the <TT>.ext</TT> and <TT>.res.ext</TT>
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
files. When <B>extract do resistance</B> is used, there is no need
to run <B>extresist</B> as a separate command. However,
<B>extresist</B> may be run prior to <B>extract</B> to set the
options that affect resistance network extraction, such as
<B>extresist threshold</B>, <B>extresist minres</B>, etc. <P>
As of magic version 8.3.653, the <B>extresist tolerance</B> option
is deprecated and has no effect on network resistance extraction.
Instead, extraction is controlled by three main options: <P>
<B>extresist threshold</B> (value in milliohms) sets a cutoff for
considering a network for detailed resistance extraction based on
the lumped resistance estimate (see above). For point-to-point
wires, the lumped resistance is approximately equal to the actual
wire resistance. For branching networks, it will generally be an
over-estimate. The default <B>threshold</B> value is set to
10 ohms. <P>
<B>extresist minres</B> (value in milliohms) sets a cutoff for
individual resistors in the detailed resistor network. Resistors
below this threshold will get pruned out of the network if the
simplification algorithm is able to remove them. The default
<B>minres</B> value is set to 1 ohm. <P>
<B>extresist mindelay</B> (value in picoseconds) sets a cutoff for
considering a network for detailed resistance extraction based on
the end-to-end delay calculated from the lumped resistance
estimate. Once the network has been extracted, the <B>mindelay</B>
value is again evaluated against a revised calculation of the
delay to determine if the extracted detailed network should be
output. The default <B>mindelay<B> value is set to 0, indicating
that only the lumped resistance threshold should be used for
determining when to extract a detailed resistance network.
<B>mindelay</B> may be used in place of <B>threshold</B>, or both
may be used together, in which case a network is only extracted
if both <B>threshold</B> and <B>mindelay</B> are exceeded. <P>
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
More details on using <B>extresist</B> can be found in
<B>magic</B> Tutorial number 8.
</BLOCKQUOTE>
<H3>Implementation Notes:</H3>
<BLOCKQUOTE>
<B>extresist</B> is implemented as a built-in command in <B>magic</B>.
</BLOCKQUOTE>
<H3>See Also:</H3>
<BLOCKQUOTE>
<A HREF=extract.html><B>extract</B></A> <BR>
<A HREF=ext2sim.html><B>ext2sim</B></A> <BR>
Updated some of the HTML command reference documentation to reflect changes just made in the "extract" and "extresist" commands.
2026-01-28 23:05:00 +01:00
<A HREF=ext2spice.html><B>ext2spice</B></A> <BR>
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
</BLOCKQUOTE>
<P><IMG SRC=graphics/line1.gif><P>
<TABLE BORDER=0>
<TR>
<TD> <A HREF=commands.html>Return to command index</A>
</TR>
</TABLE>
More work on the "extresist" code, this time to revise the way that "extresist" selects or rejects networks for resistance extraction. The "tolerance" value, which nobody (including myself) seems to understand, is now deprecated, replaced by a handful of more meaningful options: "threshold" to set a lumped resistance estimate threshold to determined whether or not to extract a network; "minres" to specify when to prune small resistors from the resulting tree network, and "mindelay" to gate the selection and output on the estimated signal propagation delay. With these settings, I am much better able to control the size and complexity of the resulting output.
2026-05-29 17:47:01 +02:00
<P><I>Last updated:</I> May 29, 2026 at 11:33am <P>
Initial commit at Tue Apr 25 08:41:48 EDT 2017 by tim on stravinsky
2017-04-25 14:41:48 +02:00
</BODY>
</HTML>