riscv
/
yosys
mirror of https://github.com/YosysHQ/yosys.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
yosys/manual/command-reference-manual.tex

3999 lines
129 KiB
TeX
Raw Blame History

% Generated using the yosys 'help -write-tex-command-reference-manual' command.
\section{abc -- use ABC for technology mapping}
\label{cmd:abc}
\begin{lstlisting}[numbers=left,frame=single]
abc [options] [selection]
This pass uses the ABC tool [1] for technology mapping of yosys's internal gate
library to a target architecture.
-exe <command>
use the specified command name instead of "yosys-abc" to execute ABC.
This can e.g. be used to call a specific version of ABC or a wrapper.
-script <file>
use the specified ABC script file instead of the default script.
if <file> starts with a plus sign (+), then the rest of the filename
string is interpreted as the command string to be passed to ABC. The
leading plus sign is removed and all commas (,) in the string are
replaced with blanks before the string is passed to ABC.
if no -script parameter is given, the following scripts are used:
for -liberty without -constr:
strash; scorr; ifraig; retime {D}; strash; dch -f; map {D}
for -liberty with -constr:
strash; scorr; ifraig; retime {D}; strash; dch -f; map {D};
buffer; upsize {D}; dnsize {D}; stime -p
for -lut:
strash; scorr; ifraig; retime; strash; dch -f; if
otherwise:
strash; scorr; ifraig; retime; strash; dch -f; map
-fast
use different default scripts that are slightly faster (at the cost
of output quality):
for -liberty without -constr:
retime {D}; map {D}
for -liberty with -constr:
retime {D}; map {D}; buffer; upsize {D}; dnsize {D}; stime -p
for -lut:
retime; if
otherwise:
retime; map
-liberty <file>
generate netlists for the specified cell library (using the liberty
file format).
-constr <file>
pass this file with timing constraints to ABC. use with -liberty.
a constr file contains two lines:
set_driving_cell <cell_name>
set_load <floating_point_number>
the set_driving_cell statement defines which cell type is assumed to
drive the primary inputs and the set_load statement sets the load in
femtofarads for each primary output.
-D <picoseconds>
set delay target. the string {D} in the default scripts above is
replaced by this option when used, and an empty string otherwise.
-lut <width>
generate netlist using luts of (max) the specified width.
-lut <w1>:<w2>
generate netlist using luts of (max) the specified width <w2>. All
luts with width <= <w1> have constant cost. for luts larger than <w1>
the area cost doubles with each additional input bit. the delay cost
is still constant for all lut widths.
-luts <cost1>,<cost2>,<cost3>,<sizeN>:<cost4-N>,..
generate netlist using luts. Use the specified costs for luts with 1,
2, 3, .. inputs.
-g type1,type2,...
Map the the specified list of gate types. Supported gates types are:
AND, NAND, OR, NOR, XOR, XNOR, MUX, AOI3, OAI3, AOI4, OAI4.
(The NOT gate is always added to this list automatically.)
-dff
also pass $_DFF_?_ and $_DFFE_??_ cells through ABC. modules with many
clock domains are automatically partitioned in clock domains and each
domain is passed through ABC independently.
-clk [!]<clock-signal-name>[,[!]<enable-signal-name>]
use only the specified clock domain. this is like -dff, but only FF
cells that belong to the specified clock domain are used.
-keepff
set the "keep" attribute on flip-flop output wires. (and thus preserve
them, for example for equivalence checking.)
-nocleanup
when this option is used, the temporary files created by this pass
are not removed. this is useful for debugging.
-showtmp
print the temp dir name in log. usually this is suppressed so that the
command output is identical across runs.
-markgroups
set a 'abcgroup' attribute on all objects created by ABC. The value of
this attribute is a unique integer for each ABC process started. This
is useful for debugging the partitioning of clock domains.
When neither -liberty nor -lut is used, the Yosys standard cell library is
loaded into ABC before the ABC script is executed.
This pass does not operate on modules with unprocessed processes in it.
(I.e. the 'proc' pass should be used first to convert processes to netlists.)
[1] http://www.eecs.berkeley.edu/~alanmi/abc/
\end{lstlisting}
\section{add -- add objects to the design}
\label{cmd:add}
\begin{lstlisting}[numbers=left,frame=single]
add <command> [selection]
This command adds objects to the design. It operates on all fully selected
modules. So e.g. 'add -wire foo' will add a wire foo to all selected modules.
add {-wire|-input|-inout|-output} <name> <width> [selection]
Add a wire (input, inout, output port) with the given name and width. The
command will fail if the object exists already and has different properties
than the object to be created.
add -global_input <name> <width> [selection]
Like 'add -input', but also connect the signal between instances of the
selected modules.
\end{lstlisting}
\section{aigmap -- map logic to and-inverter-graph circuit}
\label{cmd:aigmap}
\begin{lstlisting}[numbers=left,frame=single]
aigmap [options] [selection]
Replace all logic cells with circuits made of only $_AND_ and
$_NOT_ cells.
-nand
Enable creation of $_NAND_ cells
\end{lstlisting}
\section{alumacc -- extract ALU and MACC cells}
\label{cmd:alumacc}
\begin{lstlisting}[numbers=left,frame=single]
alumacc [selection]
This pass translates arithmetic operations like $add, $mul, $lt, etc. to $alu
and $macc cells.
\end{lstlisting}
\section{cd -- a shortcut for 'select -module <name>'}
\label{cmd:cd}
\begin{lstlisting}[numbers=left,frame=single]
cd <modname>
This is just a shortcut for 'select -module <modname>'.
cd <cellname>
When no module with the specified name is found, but there is a cell
with the specified name in the current module, then this is equivalent
to 'cd <celltype>'.
cd ..
This is just a shortcut for 'select -clear'.
\end{lstlisting}
\section{check -- check for obvious problems in the design}
\label{cmd:check}
\begin{lstlisting}[numbers=left,frame=single]
check [options] [selection]
This pass identifies the following problems in the current design:
- combinatorial loops
- two or more conflicting drivers for one wire
- used wires that do not have a driver
When called with -noinit then this command also checks for wires which have
the 'init' attribute set.
When called with -assert then the command will produce an error if any
problems are found in the current design.
\end{lstlisting}
\section{chparam -- re-evaluate modules with new parameters}
\label{cmd:chparam}
\begin{lstlisting}[numbers=left,frame=single]
chparam [ -set name value ]... [selection]
Re-evaluate the selected modules with new parameters. String values must be
passed in double quotes (").
chparam -list [selection]
List the available parameters of the selected modules.
\end{lstlisting}
\section{clean -- remove unused cells and wires}
\label{cmd:clean}
\begin{lstlisting}[numbers=left,frame=single]
clean [options] [selection]
This is identical to 'opt_clean', but less verbose.
When commands are separated using the ';;' token, this command will be executed
between the commands.
When commands are separated using the ';;;' token, this command will be executed
in -purge mode between the commands.
\end{lstlisting}
\section{connect -- create or remove connections}
\label{cmd:connect}
\begin{lstlisting}[numbers=left,frame=single]
connect [-nomap] [-nounset] -set <lhs-expr> <rhs-expr>
Create a connection. This is equivalent to adding the statement 'assign
<lhs-expr> = <rhs-expr>;' to the Verilog input. Per default, all existing
drivers for <lhs-expr> are unconnected. This can be overwritten by using
the -nounset option.
connect [-nomap] -unset <expr>
Unconnect all existing drivers for the specified expression.
connect [-nomap] -port <cell> <port> <expr>
Connect the specified cell port to the specified cell port.
Per default signal alias names are resolved and all signal names are mapped
the the signal name of the primary driver. Using the -nomap option deactivates
this behavior.
The connect command operates in one module only. Either only one module must
be selected or an active module must be set using the 'cd' command.
This command does not operate on module with processes.
\end{lstlisting}
\section{connwrappers -- replace undef values with defined constants}
\label{cmd:connwrappers}
\begin{lstlisting}[numbers=left,frame=single]
connwrappers [options] [selection]
Wrappers are used in coarse-grain synthesis to wrap cells with smaller ports
in wrapper cells with a (larger) constant port size. I.e. the upper bits
of the wrapper output are signed/unsigned bit extended. This command uses this
knowledge to rewire the inputs of the driven cells to match the output of
the driving cell.
-signed <cell_type> <port_name> <width_param>
-unsigned <cell_type> <port_name> <width_param>
consider the specified signed/unsigned wrapper output
-port <cell_type> <port_name> <width_param> <sign_param>
use the specified parameter to decide if signed or unsigned
The options -signed, -unsigned, and -port can be specified multiple times.
\end{lstlisting}
\section{copy -- copy modules in the design}
\label{cmd:copy}
\begin{lstlisting}[numbers=left,frame=single]
copy old_name new_name
Copy the specified module. Note that selection patterns are not supported
by this command.
\end{lstlisting}
\section{cover -- print code coverage counters}
\label{cmd:cover}
\begin{lstlisting}[numbers=left,frame=single]
cover [options] [pattern]
Print the code coverage counters collected using the cover() macro in the Yosys
C++ code. This is useful to figure out what parts of Yosys are utilized by a
test bench.
-q
Do not print output to the normal destination (console and/or log file)
-o file
Write output to this file, truncate if exists.
-a file
Write output to this file, append if exists.
-d dir
Write output to a newly created file in the specified directory.
When one or more pattern (shell wildcards) are specified, then only counters
matching at least one pattern are printed.
It is also possible to instruct Yosys to print the coverage counters on program
exit to a file using environment variables:
YOSYS_COVER_DIR="{dir-name}" yosys {args}
This will create a file (with an auto-generated name) in this
directory and write the coverage counters to it.
YOSYS_COVER_FILE="{file-name}" yosys {args}
This will append the coverage counters to the specified file.
Hint: Use the following AWK command to consolidate Yosys coverage files:
gawk '{ p[$3] = $1; c[$3] += $2; } END { for (i in p)
printf "%-60s %10d %s\n", p[i], c[i], i; }' {files} | sort -k3
Coverage counters are only available in Yosys for Linux.
\end{lstlisting}
\section{delete -- delete objects in the design}
\label{cmd:delete}
\begin{lstlisting}[numbers=left,frame=single]
delete [selection]
Deletes the selected objects. This will also remove entire modules, if the
whole module is selected.
delete {-input|-output|-port} [selection]
Does not delete any object but removes the input and/or output flag on the
selected wires, thus 'deleting' module ports.
\end{lstlisting}
\section{design -- save, restore and reset current design}
\label{cmd:design}
\begin{lstlisting}[numbers=left,frame=single]
design -reset
Clear the current design.
design -save <name>
Save the current design under the given name.
design -stash <name>
Save the current design under the given name and then clear the current design.
design -push
Push the current design to the stack and then clear the current design.
design -pop
Reset the current design and pop the last design from the stack.
design -load <name>
Reset the current design and load the design previously saved under the given
name.
design -copy-from <name> [-as <new_mod_name>] <selection>
Copy modules from the specified design into the current one. The selection is
evaluated in the other design.
design -copy-to <name> [-as <new_mod_name>] [selection]
Copy modules from the current design into the specified one.
\end{lstlisting}
\section{dff2dffe -- transform \$dff cells to \$dffe cells}
\label{cmd:dff2dffe}
\begin{lstlisting}[numbers=left,frame=single]
dff2dffe [options] [selection]
This pass transforms $dff cells driven by a tree of multiplexers with one or
more feedback paths to $dffe cells. It also works on gate-level cells such as
$_DFF_P_, $_DFF_N_ and $_MUX_.
-unmap
operate in the opposite direction: replace $dffe cells with combinations
of $dff and $mux cells. the options below are ignore in unmap mode.
-direct <internal_gate_type> <external_gate_type>
map directly to external gate type. <internal_gate_type> can
be any internal gate-level FF cell (except $_DFFE_??_). the
<external_gate_type> is the cell type name for a cell with an
identical interface to the <internal_gate_type>, except it
also has an high-active enable port 'E'.
Usually <external_gate_type> is an intermediate cell type
that is then translated to the final type using 'techmap'.
-direct-match <pattern>
like -direct for all DFF cell types matching the expression.
this will use $__DFFE_* as <external_gate_type> matching the
internal gate type $_DFF_*_, except for $_DFF_[NP]_, which is
converted to $_DFFE_[NP]_.
\end{lstlisting}
\section{dffinit -- set INIT param on FF cells}
\label{cmd:dffinit}
\begin{lstlisting}[numbers=left,frame=single]
dffinit [options] [selection]
This pass sets an FF cell parameter to the the initial value of the net it
drives. (This is primarily used in FPGA flows.)
-ff <cell_name> <output_port> <init_param>
operate on the specified cell type. this option can be used
multiple times.
\end{lstlisting}
\section{dfflibmap -- technology mapping of flip-flops}
\label{cmd:dfflibmap}
\begin{lstlisting}[numbers=left,frame=single]
dfflibmap [-prepare] -liberty <file> [selection]
Map internal flip-flop cells to the flip-flop cells in the technology
library specified in the given liberty file.
This pass may add inverters as needed. Therefore it is recommended to
first run this pass and then map the logic paths to the target technology.
When called with -prepare, this command will convert the internal FF cells
to the internal cell types that best match the cells found in the given
liberty file.
\end{lstlisting}
\section{dffsr2dff -- convert DFFSR cells to simpler FF cell types}
\label{cmd:dffsr2dff}
\begin{lstlisting}[numbers=left,frame=single]
dffsr2dff [options] [selection]
This pass converts DFFSR cells ($dffsr, $_DFFSR_???_) and ADFF cells ($adff,
$_DFF_???_) to simpler FF cell types when any of the set/reset inputs is unused.
\end{lstlisting}
\section{dump -- print parts of the design in ilang format}
\label{cmd:dump}
\begin{lstlisting}[numbers=left,frame=single]
dump [options] [selection]
Write the selected parts of the design to the console or specified file in
ilang format.
-m
also dump the module headers, even if only parts of a single
module is selected
-n
only dump the module headers if the entire module is selected
-o <filename>
write to the specified file.
-a <filename>
like -outfile but append instead of overwrite
\end{lstlisting}
\section{echo -- turning echoing back of commands on and off}
\label{cmd:echo}
\begin{lstlisting}[numbers=left,frame=single]
echo on
Print all commands to log before executing them.
echo off
Do not print all commands to log before executing them. (default)
\end{lstlisting}
\section{edgetypes -- list all types of edges in selection}
\label{cmd:edgetypes}
\begin{lstlisting}[numbers=left,frame=single]
edgetypes [options] [selection]
This command lists all unique types of 'edges' found in the selection. An 'edge'
is a 4-tuple of source and sink cell type and port name.
\end{lstlisting}
\section{equiv\_add -- add a \$equiv cell}
\label{cmd:equiv_add}
\begin{lstlisting}[numbers=left,frame=single]
equiv_add [-try] gold_sig gate_sig
This command adds an $equiv cell for the specified signals.
equiv_add [-try] -cell gold_cell gate_cell
This command adds $equiv cells for the ports of the specified cells.
\end{lstlisting}
\section{equiv\_induct -- proving \$equiv cells using temporal induction}
\label{cmd:equiv_induct}
\begin{lstlisting}[numbers=left,frame=single]
equiv_induct [options] [selection]
Uses a version of temporal induction to prove $equiv cells.
Only selected $equiv cells are proven and only selected cells are used to
perform the proof.
-undef
enable modelling of undef states
-seq <N>
the max. number of time steps to be considered (default = 4)
This command is very effective in proving complex sequential circuits, when
the internal state of the circuit quickly propagates to $equiv cells.
However, this command uses a weak definition of 'equivalence': This command
proves that the two circuits will not diverge after they produce equal
outputs (observable points via $equiv) for at least <N> cycles (the <N>
specified via -seq).
Combined with simulation this is very powerful because simulation can give
you confidence that the circuits start out synced for at least <N> cycles
after reset.
\end{lstlisting}
\section{equiv\_make -- prepare a circuit for equivalence checking}
\label{cmd:equiv_make}
\begin{lstlisting}[numbers=left,frame=single]
equiv_make [options] gold_module gate_module equiv_module
This creates a module annotated with $equiv cells from two presumably
equivalent modules. Use commands such as 'equiv_simple' and 'equiv_status'
to work with the created equivalent checking module.
-inames
Also match cells and wires with $... names.
-blacklist <file>
Do not match cells or signals that match the names in the file.
-encfile <file>
Match FSM encodings using the description from the file.
See 'help fsm_recode' for details.
Note: The circuit created by this command is not a miter (with something like
a trigger output), but instead uses $equiv cells to encode the equivalence
checking problem. Use 'miter -equiv' if you want to create a miter circuit.
\end{lstlisting}
\section{equiv\_mark -- mark equivalence checking regions}
\label{cmd:equiv_mark}
\begin{lstlisting}[numbers=left,frame=single]
equiv_mark [options] [selection]
This command marks the regions in an equivalence checking module. Region 0 is
the proven part of the circuit. Regions with higher numbers are connected
unproven subcricuits. The integer attribute 'equiv_region' is set on all
wires and cells.
\end{lstlisting}
\section{equiv\_miter -- extract miter from equiv circuit}
\label{cmd:equiv_miter}
\begin{lstlisting}[numbers=left,frame=single]
equiv_miter [options] miter_module [selection]
This creates a miter module for further analysis of the selected $equiv cells.
-trigger
Create a trigger output
-cmp
Create cmp_* outputs for individual unproven $equiv cells
-assert
Create a $assert cell for each unproven $equiv cell
-undef
Create compare logic that handles undefs correctly
\end{lstlisting}
\section{equiv\_purge -- purge equivalence checking module}
\label{cmd:equiv_purge}
\begin{lstlisting}[numbers=left,frame=single]
equiv_purge [options] [selection]
This command removes the proven part of an equivalence checking module, leaving
only the unproven segments in the design. This will also remove and add module
ports as needed.
\end{lstlisting}
\section{equiv\_remove -- remove \$equiv cells}
\label{cmd:equiv_remove}
\begin{lstlisting}[numbers=left,frame=single]
equiv_remove [options] [selection]
This command removes the selected $equiv cells. If neither -gold nor -gate is
used then only proven cells are removed.
-gold
keep gold circuit
-gate
keep gate circuit
\end{lstlisting}
\section{equiv\_simple -- try proving simple \$equiv instances}
\label{cmd:equiv_simple}
\begin{lstlisting}[numbers=left,frame=single]
equiv_simple [options] [selection]
This command tries to prove $equiv cells using a simple direct SAT approach.
-v
verbose output
-undef
enable modelling of undef states
-nogroup
disabling grouping of $equiv cells by output wire
-seq <N>
the max. number of time steps to be considered (default = 1)
\end{lstlisting}
\section{equiv\_status -- print status of equivalent checking module}
\label{cmd:equiv_status}
\begin{lstlisting}[numbers=left,frame=single]
equiv_status [options] [selection]
This command prints status information for all selected $equiv cells.
-assert
produce an error if any unproven $equiv cell is found
\end{lstlisting}
\section{equiv\_struct -- structural equivalence checking}
\label{cmd:equiv_struct}
\begin{lstlisting}[numbers=left,frame=single]
equiv_struct [options] [selection]
This command adds additional $equiv cells based on the assumption that the
gold and gate circuit are structurally equivalent. Note that this can introduce
bad $equiv cells in cases where the netlists are not structurally equivalent,
for example when analyzing circuits with cells with commutative inputs. This
command will also de-duplicate gates.
-fwd
by default this command performans forward sweeps until nothing can
be merged by forwards sweeps, then backward sweeps until forward
sweeps are effective again. with this option set only forward sweeps
are performed.
-fwonly <cell_type>
add the specified cell type to the list of cell types that are only
merged in forward sweeps and never in backward sweeps. $equiv is in
this list automatically.
-icells
by default, the internal RTL and gate cell types are ignored. add
this option to also process those cell types with this command.
-maxiter <N>
maximum number of iterations to run before aborting
\end{lstlisting}
\section{eval -- evaluate the circuit given an input}
\label{cmd:eval}
\begin{lstlisting}[numbers=left,frame=single]
eval [options] [selection]
This command evaluates the value of a signal given the value of all required
inputs.
-set <signal> <value>
set the specified signal to the specified value.
-set-undef
set all unspecified source signals to undef (x)
-table <signal>
create a truth table using the specified input signals
-show <signal>
show the value for the specified signal. if no -show option is passed
then all output ports of the current module are used.
\end{lstlisting}
\section{expose -- convert internal signals to module ports}
\label{cmd:expose}
\begin{lstlisting}[numbers=left,frame=single]
expose [options] [selection]
This command exposes all selected internal signals of a module as additional
outputs.
-dff
only consider wires that are directly driven by register cell.
-cut
when exposing a wire, create an input/output pair and cut the internal
signal path at that wire.
-shared
only expose those signals that are shared among the selected modules.
this is useful for preparing modules for equivalence checking.
-evert
also turn connections to instances of other modules to additional
inputs and outputs and remove the module instances.
-evert-dff
turn flip-flops to sets of inputs and outputs.
-sep <separator>
when creating new wire/port names, the original object name is suffixed
with this separator (default: '.') and the port name or a type
designator for the exposed signal.
\end{lstlisting}
\section{extract -- find subcircuits and replace them with cells}
\label{cmd:extract}
\begin{lstlisting}[numbers=left,frame=single]
extract -map <map_file> [options] [selection]
extract -mine <out_file> [options] [selection]
This pass looks for subcircuits that are isomorphic to any of the modules
in the given map file and replaces them with instances of this modules. The
map file can be a Verilog source file (*.v) or an ilang file (*.il).
-map <map_file>
use the modules in this file as reference. This option can be used
multiple times.
-map %<design-name>
use the modules in this in-memory design as reference. This option can
be used multiple times.
-verbose
print debug output while analyzing
-constports
also find instances with constant drivers. this may be much
slower than the normal operation.
-nodefaultswaps
normally builtin port swapping rules for internal cells are used per
default. This turns that off, so e.g. 'a^b' does not match 'b^a'
when this option is used.
-compat <needle_type> <haystack_type>
Per default, the cells in the map file (needle) must have the
type as the cells in the active design (haystack). This option
can be used to register additional pairs of types that should
match. This option can be used multiple times.
-swap <needle_type> <port1>,<port2>[,...]
Register a set of swappable ports for a needle cell type.
This option can be used multiple times.
-perm <needle_type> <port1>,<port2>[,...] <portA>,<portB>[,...]
Register a valid permutation of swappable ports for a needle
cell type. This option can be used multiple times.
-cell_attr <attribute_name>
Attributes on cells with the given name must match.
-wire_attr <attribute_name>
Attributes on wires with the given name must match.
-ignore_parameters
Do not use parameters when matching cells.
-ignore_param <cell_type> <parameter_name>
Do not use this parameter when matching cells.
This pass does not operate on modules with unprocessed processes in it.
(I.e. the 'proc' pass should be used first to convert processes to netlists.)
This pass can also be used for mining for frequent subcircuits. In this mode
the following options are to be used instead of the -map option.
-mine <out_file>
mine for frequent subcircuits and write them to the given ilang file
-mine_cells_span <min> <max>
only mine for subcircuits with the specified number of cells
default value: 3 5
-mine_min_freq <num>
only mine for subcircuits with at least the specified number of matches
default value: 10
-mine_limit_matches_per_module <num>
when calculating the number of matches for a subcircuit, don't count
more than the specified number of matches per module
-mine_max_fanout <num>
don't consider internal signals with more than <num> connections
The modules in the map file may have the attribute 'extract_order' set to an
integer value. Then this value is used to determine the order in which the pass
tries to map the modules to the design (ascending, default value is 0).
See 'help techmap' for a pass that does the opposite thing.
\end{lstlisting}
\section{flatten -- flatten design}
\label{cmd:flatten}
\begin{lstlisting}[numbers=left,frame=single]
flatten [selection]
This pass flattens the design by replacing cells by their implementation. This
pass is very similar to the 'techmap' pass. The only difference is that this
pass is using the current design as mapping library.
Cells and/or modules with the 'keep_hierarchy' attribute set will not be
flattened by this command.
\end{lstlisting}
\section{freduce -- perform functional reduction}
\label{cmd:freduce}
\begin{lstlisting}[numbers=left,frame=single]
freduce [options] [selection]
This pass performs functional reduction in the circuit. I.e. if two nodes are
equivalent, they are merged to one node and one of the redundant drivers is
disconnected. A subsequent call to 'clean' will remove the redundant drivers.
-v, -vv
enable verbose or very verbose output
-inv
enable explicit handling of inverted signals
-stop <n>
stop after <n> reduction operations. this is mostly used for
debugging the freduce command itself.
-dump <prefix>
dump the design to <prefix>_<module>_<num>.il after each reduction
operation. this is mostly used for debugging the freduce command.
This pass is undef-aware, i.e. it considers don't-care values for detecting
equivalent nodes.
All selected wires are considered for rewiring. The selected cells cover the
circuit that is analyzed.
\end{lstlisting}
\section{fsm -- extract and optimize finite state machines}
\label{cmd:fsm}
\begin{lstlisting}[numbers=left,frame=single]
fsm [options] [selection]
This pass calls all the other fsm_* passes in a useful order. This performs
FSM extraction and optimization. It also calls opt_clean as needed:
fsm_detect unless got option -nodetect
fsm_extract
fsm_opt
opt_clean
fsm_opt
fsm_expand if got option -expand
opt_clean if got option -expand
fsm_opt if got option -expand
fsm_recode unless got option -norecode
fsm_info
fsm_export if got option -export
fsm_map unless got option -nomap
Options:
-expand, -norecode, -export, -nomap
enable or disable passes as indicated above
-encoding type
-fm_set_fsm_file file
-encfile file
passed through to fsm_recode pass
\end{lstlisting}
\section{fsm\_detect -- finding FSMs in design}
\label{cmd:fsm_detect}
\begin{lstlisting}[numbers=left,frame=single]
fsm_detect [selection]
This pass detects finite state machines by identifying the state signal.
The state signal is then marked by setting the attribute 'fsm_encoding'
on the state signal to "auto".
Existing 'fsm_encoding' attributes are not changed by this pass.
Signals can be protected from being detected by this pass by setting the
'fsm_encoding' attribute to "none".
\end{lstlisting}
\section{fsm\_expand -- expand FSM cells by merging logic into it}
\label{cmd:fsm_expand}
\begin{lstlisting}[numbers=left,frame=single]
fsm_expand [selection]
The fsm_extract pass is conservative about the cells that belong to a finite
state machine. This pass can be used to merge additional auxiliary gates into
the finite state machine.
\end{lstlisting}
\section{fsm\_export -- exporting FSMs to KISS2 files}
\label{cmd:fsm_export}
\begin{lstlisting}[numbers=left,frame=single]
fsm_export [-noauto] [-o filename] [-origenc] [selection]
This pass creates a KISS2 file for every selected FSM. For FSMs with the
'fsm_export' attribute set, the attribute value is used as filename, otherwise
the module and cell name is used as filename. If the parameter '-o' is given,
the first exported FSM is written to the specified filename. This overwrites
the setting as specified with the 'fsm_export' attribute. All other FSMs are
exported to the default name as mentioned above.
-noauto
only export FSMs that have the 'fsm_export' attribute set
-o filename
filename of the first exported FSM
-origenc
use binary state encoding as state names instead of s0, s1, ...
\end{lstlisting}
\section{fsm\_extract -- extracting FSMs in design}
\label{cmd:fsm_extract}
\begin{lstlisting}[numbers=left,frame=single]
fsm_extract [selection]
This pass operates on all signals marked as FSM state signals using the
'fsm_encoding' attribute. It consumes the logic that creates the state signal
and uses the state signal to generate control signal and replaces it with an
FSM cell.
The generated FSM cell still generates the original state signal with its
original encoding. The 'fsm_opt' pass can be used in combination with the
'opt_clean' pass to eliminate this signal.
\end{lstlisting}
\section{fsm\_info -- print information on finite state machines}
\label{cmd:fsm_info}
\begin{lstlisting}[numbers=left,frame=single]
fsm_info [selection]
This pass dumps all internal information on FSM cells. It can be useful for
analyzing the synthesis process and is called automatically by the 'fsm'
pass so that this information is included in the synthesis log file.
\end{lstlisting}
\section{fsm\_map -- mapping FSMs to basic logic}
\label{cmd:fsm_map}
\begin{lstlisting}[numbers=left,frame=single]
fsm_map [selection]
This pass translates FSM cells to flip-flops and logic.
\end{lstlisting}
\section{fsm\_opt -- optimize finite state machines}
\label{cmd:fsm_opt}
\begin{lstlisting}[numbers=left,frame=single]
fsm_opt [selection]
This pass optimizes FSM cells. It detects which output signals are actually
not used and removes them from the FSM. This pass is usually used in
combination with the 'opt_clean' pass (see also 'help fsm').
\end{lstlisting}
\section{fsm\_recode -- recoding finite state machines}
\label{cmd:fsm_recode}
\begin{lstlisting}[numbers=left,frame=single]
fsm_recode [options] [selection]
This pass reassign the state encodings for FSM cells. At the moment only
one-hot encoding and binary encoding is supported.
-encoding <type>
specify the encoding scheme used for FSMs without the
'fsm_encoding' attribute or with the attribute set to `auto'.
-fm_set_fsm_file <file>
generate a file containing the mapping from old to new FSM encoding
in form of Synopsys Formality set_fsm_* commands.
-encfile <file>
write the mappings from old to new FSM encoding to a file in the
following format:
.fsm <module_name> <state_signal>
.map <old_bitpattern> <new_bitpattern>
\end{lstlisting}
\section{help -- display help messages}
\label{cmd:help}
\begin{lstlisting}[numbers=left,frame=single]
help ................ list all commands
help <command> ...... print help message for given command
help -all ........... print complete command reference
help -cells .......... list all cell types
help <celltype> ..... print help message for given cell type
help <celltype>+ .... print verilog code for given cell type
\end{lstlisting}
\section{hierarchy -- check, expand and clean up design hierarchy}
\label{cmd:hierarchy}
\begin{lstlisting}[numbers=left,frame=single]
hierarchy [-check] [-top <module>]
hierarchy -generate <cell-types> <port-decls>
In parametric designs, a module might exists in several variations with
different parameter values. This pass looks at all modules in the current
design an re-runs the language frontends for the parametric modules as
needed.
-check
also check the design hierarchy. this generates an error when
an unknown module is used as cell type.
-purge_lib
by default the hierarchy command will not remove library (blackbox)
modules. use this option to also remove unused blackbox modules.
-libdir <directory>
search for files named <module_name>.v in the specified directory
for unknown modules and automatically run read_verilog for each
unknown module.
-keep_positionals
per default this pass also converts positional arguments in cells
to arguments using port names. this option disables this behavior.
-nokeep_asserts
per default this pass sets the "keep" attribute on all modules
that directly or indirectly contain one or more $assert cells. this
option disables this behavior.
-top <module>
use the specified top module to built a design hierarchy. modules
outside this tree (unused modules) are removed.
when the -top option is used, the 'top' attribute will be set on the
specified top module. otherwise a module with the 'top' attribute set
will implicitly be used as top module, if such a module exists.
-auto-top
automatically determine the top of the design hierarchy and mark it.
In -generate mode this pass generates blackbox modules for the given cell
types (wildcards supported). For this the design is searched for cells that
match the given types and then the given port declarations are used to
determine the direction of the ports. The syntax for a port declaration is:
{i|o|io}[@<num>]:<portname>
Input ports are specified with the 'i' prefix, output ports with the 'o'
prefix and inout ports with the 'io' prefix. The optional <num> specifies
the position of the port in the parameter list (needed when instantiated
using positional arguments). When <num> is not specified, the <portname> can
also contain wildcard characters.
This pass ignores the current selection and always operates on all modules
in the current design.
\end{lstlisting}
\section{hilomap -- technology mapping of constant hi- and/or lo-drivers}
\label{cmd:hilomap}
\begin{lstlisting}[numbers=left,frame=single]
hilomap [options] [selection]
Map constants to 'tielo' and 'tiehi' driver cells.
-hicell <celltype> <portname>
Replace constant hi bits with this cell.
-locell <celltype> <portname>
Replace constant lo bits with this cell.
-singleton
Create only one hi/lo cell and connect all constant bits
to that cell. Per default a separate cell is created for
each constant bit.
\end{lstlisting}
\section{history -- show last interactive commands}
\label{cmd:history}
\begin{lstlisting}[numbers=left,frame=single]
history
This command prints all commands in the shell history buffer. This are
all commands executed in an interactive session, but not the commands
from executed scripts.
\end{lstlisting}
\section{ice40\_ffinit -- iCE40: handle FF init values}
\label{cmd:ice40_ffinit}
\begin{lstlisting}[numbers=left,frame=single]
ice40_ffinit [options] [selection]
Remove zero init values for FF output signals. Add inverters to implement
nonzero init values.
\end{lstlisting}
\section{ice40\_ffssr -- iCE40: merge synchronous set/reset into FF cells}
\label{cmd:ice40_ffssr}
\begin{lstlisting}[numbers=left,frame=single]
ice40_ffssr [options] [selection]
Merge synchronous set/reset $_MUX_ cells into iCE40 FFs.
\end{lstlisting}
\section{ice40\_opt -- iCE40: perform simple optimizations}
\label{cmd:ice40_opt}
\begin{lstlisting}[numbers=left,frame=single]
ice40_opt [options] [selection]
This command executes the following script:
do
<ice40 specific optimizations>
opt_const -mux_undef -undriven [-full]
opt_share
opt_rmdff
opt_clean
while <changed design>
\end{lstlisting}
\section{iopadmap -- technology mapping of i/o pads (or buffers)}
\label{cmd:iopadmap}
\begin{lstlisting}[numbers=left,frame=single]
iopadmap [options] [selection]
Map module inputs/outputs to PAD cells from a library. This pass
can only map to very simple PAD cells. Use 'techmap' to further map
the resulting cells to more sophisticated PAD cells.
-inpad <celltype> <portname>[:<portname>]
Map module input ports to the given cell type with the
given output port name. if a 2nd portname is given, the
signal is passed through the pad call, using the 2nd
portname as input.
-outpad <celltype> <portname>[:<portname>]
-inoutpad <celltype> <portname>[:<portname>]
Similar to -inpad, but for output and inout ports.
-widthparam <param_name>
Use the specified parameter name to set the port width.
-nameparam <param_name>
Use the specified parameter to set the port name.
-bits
create individual bit-wide buffers even for ports that
are wider. (the default behavior is to create word-wide
buffers using -widthparam to set the word size on the cell.)
\end{lstlisting}
\section{json -- write design in JSON format}
\label{cmd:json}
\begin{lstlisting}[numbers=left,frame=single]
json [options] [selection]
Write a JSON netlist of all selected objects.
-o <filename>
write to the specified file.
-aig
also include AIG models for the different gate types
See 'help write_json' for a description of the JSON format used.
\end{lstlisting}
\section{log -- print text and log files}
\label{cmd:log}
\begin{lstlisting}[numbers=left,frame=single]
log string
Print the given string to the screen and/or the log file. This is useful for TCL
scripts, because the TCL command "puts" only goes to stdout but not to
logfiles.
-stdout
Print the output to stdout too. This is useful when all Yosys is executed
with a script and the -q (quiet operation) argument to notify the user.
-stderr
Print the output to stderr too.
-nolog
Don't use the internal log() command. Use either -stdout or -stderr,
otherwise no output will be generated at all.
-n
do not append a newline
\end{lstlisting}
\section{ls -- list modules or objects in modules}
\label{cmd:ls}
\begin{lstlisting}[numbers=left,frame=single]
ls [selection]
When no active module is selected, this prints a list of modules.
When an active module is selected, this prints a list of objects in the module.
\end{lstlisting}
\section{lut2mux -- convert \$lut to \$\_MUX\_}
\label{cmd:lut2mux}
\begin{lstlisting}[numbers=left,frame=single]
lut2mux [options] [selection]
This pass converts $lut cells to $_MUX_ gates.
\end{lstlisting}
\section{maccmap -- mapping macc cells}
\label{cmd:maccmap}
\begin{lstlisting}[numbers=left,frame=single]
maccmap [-unmap] [selection]
This pass maps $macc cells to yosys $fa and $alu cells. When the -unmap option
is used then the $macc cell is mapped to $add, $sub, etc. cells instead.
\end{lstlisting}
\section{memory -- translate memories to basic cells}
\label{cmd:memory}
\begin{lstlisting}[numbers=left,frame=single]
memory [-nomap] [-nordff] [-bram <bram_rules>] [selection]
This pass calls all the other memory_* passes in a useful order:
memory_dff [-nordff]
opt_clean
memory_share
opt_clean
memory_collect
memory_bram -rules <bram_rules> (when called with -bram)
memory_map (skipped if called with -nomap)
This converts memories to word-wide DFFs and address decoders
or multiport memory blocks if called with the -nomap option.
\end{lstlisting}
\section{memory\_bram -- map memories to block rams}
\label{cmd:memory_bram}
\begin{lstlisting}[numbers=left,frame=single]
memory_bram -rules <rule_file> [selection]
This pass converts the multi-port $mem memory cells into block ram instances.
The given rules file describes the available resources and how they should be
used.
The rules file contains a set of block ram description and a sequence of match
rules. A block ram description looks like this:
bram RAMB1024X32 # name of BRAM cell
init 1 # set to '1' if BRAM can be initialized
abits 10 # number of address bits
dbits 32 # number of data bits
groups 2 # number of port groups
ports 1 1 # number of ports in each group
wrmode 1 0 # set to '1' if this groups is write ports
enable 4 1 # number of enable bits
transp 0 2 # transparent (for read ports)
clocks 1 2 # clock configuration
clkpol 2 2 # clock polarity configuration
endbram
For the option 'transp' the value 0 means non-transparent, 1 means transparent
and a value greater than 1 means configurable. All groups with the same
value greater than 1 share the same configuration bit.
For the option 'clocks' the value 0 means non-clocked, and a value greater
than 0 means clocked. All groups with the same value share the same clock
signal.
For the option 'clkpol' the value 0 means negative edge, 1 means positive edge
and a value greater than 1 means configurable. All groups with the same value
greater than 1 share the same configuration bit.
Using the same bram name in different bram blocks will create different variants
of the bram. Verilog configuration parameters for the bram are created as needed.
It is also possible to create variants by repeating statements in the bram block
and appending '@<label>' to the individual statements.
A match rule looks like this:
match RAMB1024X32
max waste 16384 # only use this bram if <= 16k ram bits are unused
min efficiency 80 # only use this bram if efficiency is at least 80%
endmatch
It is possible to match against the following values with min/max rules:
words ........ number of words in memory in design
abits ........ number of address bits on memory in design
dbits ........ number of data bits on memory in design
wports ....... number of write ports on memory in design
rports ....... number of read ports on memory in design
ports ........ number of ports on memory in design
bits ......... number of bits in memory in design
dups .......... number of duplications for more read ports
awaste ....... number of unused address slots for this match
dwaste ....... number of unused data bits for this match
bwaste ....... number of unused bram bits for this match
waste ........ total number of unused bram bits (bwaste*dups)
efficiency ... total percentage of used and non-duplicated bits
acells ....... number of cells in 'address-direction'
dcells ....... number of cells in 'data-direction'
cells ........ total number of cells (acells*dcells*dups)
The interface for the created bram instances is derived from the bram
description. Use 'techmap' to convert the created bram instances into
instances of the actual bram cells of your target architecture.
A match containing the command 'or_next_if_better' is only used if it
has a higher efficiency than the next match (and the one after that if
the next also has 'or_next_if_better' set, and so forth).
A match containing the command 'make_transp' will add external circuitry
to simulate 'transparent read', if necessary.
A match containing the command 'make_outreg' will add external flip-flops
to implement synchronous read ports, if necessary.
A match containing the command 'shuffle_enable A' will re-organize
the data bits to accommodate the enable pattern of port A.
\end{lstlisting}
\section{memory\_collect -- creating multi-port memory cells}
\label{cmd:memory_collect}
\begin{lstlisting}[numbers=left,frame=single]
memory_collect [selection]
This pass collects memories and memory ports and creates generic multiport
memory cells.
\end{lstlisting}
\section{memory\_dff -- merge input/output DFFs into memories}
\label{cmd:memory_dff}
\begin{lstlisting}[numbers=left,frame=single]
memory_dff [options] [selection]
This pass detects DFFs at memory ports and merges them into the memory port.
I.e. it consumes an asynchronous memory port and the flip-flops at its
interface and yields a synchronous memory port.
-nordfff
do not merge registers on read ports
\end{lstlisting}
\section{memory\_map -- translate multiport memories to basic cells}
\label{cmd:memory_map}
\begin{lstlisting}[numbers=left,frame=single]
memory_map [selection]
This pass converts multiport memory cells as generated by the memory_collect
pass to word-wide DFFs and address decoders.
\end{lstlisting}
\section{memory\_share -- consolidate memory ports}
\label{cmd:memory_share}
\begin{lstlisting}[numbers=left,frame=single]
memory_share [selection]
This pass merges share-able memory ports into single memory ports.
The following methods are used to consolidate the number of memory ports:
- When write ports are connected to async read ports accessing the same
address, then this feedback path is converted to a write port with
byte/part enable signals.
- When multiple write ports access the same address then this is converted
to a single write port with a more complex data and/or enable logic path.
- When multiple write ports are never accessed at the same time (a SAT
solver is used to determine this), then the ports are merged into a single
write port.
Note that in addition to the algorithms implemented in this pass, the $memrd
and $memwr cells are also subject to generic resource sharing passes (and other
optimizations) such as opt_share.
\end{lstlisting}
\section{memory\_unpack -- unpack multi-port memory cells}
\label{cmd:memory_unpack}
\begin{lstlisting}[numbers=left,frame=single]
memory_unpack [selection]
This pass converts the multi-port $mem memory cells into individual $memrd and
$memwr cells. It is the counterpart to the memory_collect pass.
\end{lstlisting}
\section{miter -- automatically create a miter circuit}
\label{cmd:miter}
\begin{lstlisting}[numbers=left,frame=single]
miter -equiv [options] gold_name gate_name miter_name
Creates a miter circuit for equivalence checking. The gold- and gate- modules
must have the same interfaces. The miter circuit will have all inputs of the
two source modules, prefixed with 'in_'. The miter circuit has a 'trigger'
output that goes high if an output mismatch between the two source modules is
detected.
-ignore_gold_x
a undef (x) bit in the gold module output will match any value in
the gate module output.
-make_outputs
also route the gold- and gate-outputs to 'gold_*' and 'gate_*' outputs
on the miter circuit.
-make_outcmp
also create a cmp_* output for each gold/gate output pair.
-make_assert
also create an 'assert' cell that checks if trigger is always low.
-flatten
call 'flatten; opt_const -keepdc -undriven;;' on the miter circuit.
miter -assert [options] module [miter_name]
Creates a miter circuit for property checking. All input ports are kept,
output ports are discarded. An additional output 'trigger' is created that
goes high when an assert is violated. Without a miter_name, the existing
module is modified.
-make_outputs
keep module output ports.
-flatten
call 'flatten; opt_const -keepdc -undriven;;' on the miter circuit.
\end{lstlisting}
\section{muxcover -- cover trees of MUX cells with wider MUXes}
\label{cmd:muxcover}
\begin{lstlisting}[numbers=left,frame=single]
muxcover [options] [selection]
Cover trees of $_MUX_ cells with $_MUX{4,8,16}_ cells
-mux4, -mux8, -mux16
Use the specified types of MUXes. If none of those options are used,
the effect is the same as if all of them where used.
-nodecode
Do not insert decoder logic. This reduces the number of possible
substitutions, but guarantees that the resulting circuit is not
less efficient than the original circuit.
\end{lstlisting}
\section{nlutmap -- map to LUTs of different sizes}
\label{cmd:nlutmap}
\begin{lstlisting}[numbers=left,frame=single]
nlutmap [options] [selection]
This pass uses successive calls to 'abc' to map to an architecture. That
provides a small number of differently sized LUTs.
-luts N_1,N_2,N_3,...
The number of LUTs with 1, 2, 3, ... inputs that are
available in the target architecture.
Excess logic that does not fit into the specified LUTs is mapped back
to generic logic gates ($_AND_, etc.).
\end{lstlisting}
\section{opt -- perform simple optimizations}
\label{cmd:opt}
\begin{lstlisting}[numbers=left,frame=single]
opt [options] [selection]
This pass calls all the other opt_* passes in a useful order. This performs
a series of trivial optimizations and cleanups. This pass executes the other
passes in the following order:
opt_const [-mux_undef] [-mux_bool] [-undriven] [-clkinv] [-fine] [-full] [-keepdc]
opt_share [-share_all] -nomux
do
opt_muxtree
opt_reduce [-fine] [-full]
opt_share [-share_all]
opt_rmdff
opt_clean [-purge]
opt_const [-mux_undef] [-mux_bool] [-undriven] [-clkinv] [-fine] [-full] [-keepdc]
while <changed design>
When called with -fast the following script is used instead:
do
opt_const [-mux_undef] [-mux_bool] [-undriven] [-clkinv] [-fine] [-full] [-keepdc]
opt_share [-share_all]
opt_rmdff
opt_clean [-purge]
while <changed design in opt_rmdff>
Note: Options in square brackets (such as [-keepdc]) are passed through to
the opt_* commands when given to 'opt'.
\end{lstlisting}
\section{opt\_clean -- remove unused cells and wires}
\label{cmd:opt_clean}
\begin{lstlisting}[numbers=left,frame=single]
opt_clean [options] [selection]
This pass identifies wires and cells that are unused and removes them. Other
passes often remove cells but leave the wires in the design or reconnect the
wires but leave the old cells in the design. This pass can be used to clean up
after the passes that do the actual work.
This pass only operates on completely selected modules without processes.
-purge
also remove internal nets if they have a public name
\end{lstlisting}
\section{opt\_const -- perform const folding}
\label{cmd:opt_const}
\begin{lstlisting}[numbers=left,frame=single]
opt_const [options] [selection]
This pass performs const folding on internal cell types with constant inputs.
-mux_undef
remove 'undef' inputs from $mux, $pmux and $_MUX_ cells
-mux_bool
replace $mux cells with inverters or buffers when possible
-undriven
replace undriven nets with undef (x) constants
-clkinv
optimize clock inverters by changing FF types
-fine
perform fine-grain optimizations
-full
alias for -mux_undef -mux_bool -undriven -fine
-keepdc
some optimizations change the behavior of the circuit with respect to
don't-care bits. for example in 'a+0' a single x-bit in 'a' will cause
all result bits to be set to x. this behavior changes when 'a+0' is
replaced by 'a'. the -keepdc option disables all such optimizations.
\end{lstlisting}
\section{opt\_muxtree -- eliminate dead trees in multiplexer trees}
\label{cmd:opt_muxtree}
\begin{lstlisting}[numbers=left,frame=single]
opt_muxtree [selection]
This pass analyzes the control signals for the multiplexer trees in the design
and identifies inputs that can never be active. It then removes this dead
branches from the multiplexer trees.
This pass only operates on completely selected modules without processes.
\end{lstlisting}
\section{opt\_reduce -- simplify large MUXes and AND/OR gates}
\label{cmd:opt_reduce}
\begin{lstlisting}[numbers=left,frame=single]
opt_reduce [options] [selection]
This pass performs two interlinked optimizations:
1. it consolidates trees of large AND gates or OR gates and eliminates
duplicated inputs.
2. it identifies duplicated inputs to MUXes and replaces them with a single
input with the original control signals OR'ed together.
-fine
perform fine-grain optimizations
-full
alias for -fine
\end{lstlisting}
\section{opt\_rmdff -- remove DFFs with constant inputs}
\label{cmd:opt_rmdff}
\begin{lstlisting}[numbers=left,frame=single]
opt_rmdff [selection]
This pass identifies flip-flops with constant inputs and replaces them with
a constant driver.
\end{lstlisting}
\section{opt\_share -- consolidate identical cells}
\label{cmd:opt_share}
\begin{lstlisting}[numbers=left,frame=single]
opt_share [options] [selection]
This pass identifies cells with identical type and input signals. Such cells
are then merged to one cell.
-nomux
Do not merge MUX cells.
-share_all
Operate on all cell types, not just built-in types.
\end{lstlisting}
\section{plugin -- load and list loaded plugins}
\label{cmd:plugin}
\begin{lstlisting}[numbers=left,frame=single]
plugin [options]
Load and list loaded plugins.
-i <plugin_filename>
Load (install) the specified plugin.
-a <alias_name>
Register the specified alias name for the loaded plugin
-l
List loaded plugins
\end{lstlisting}
\section{pmuxtree -- transform \$pmux cells to trees of \$mux cells}
\label{cmd:pmuxtree}
\begin{lstlisting}[numbers=left,frame=single]
pmuxtree [options] [selection]
This pass transforms $pmux cells to a trees of $mux cells.
\end{lstlisting}
\section{prep -- generic synthesis script}
\label{cmd:prep}
\begin{lstlisting}[numbers=left,frame=single]
prep [options]
This command runs a conservative RTL synthesis. A typical application for this
is the preparation stage of a verification flow. This command does not operate
on partly selected designs.
-top <module>
use the specified module as top module (default='top')
-nordff
passed to 'memory_dff'. prohibits merging of FFs into memory read ports
-run <from_label>[:<to_label>]
only run the commands between the labels (see below). an empty
from label is synonymous to 'begin', and empty to label is
synonymous to the end of the command list.
The following commands are executed by this synthesis command:
begin:
hierarchy -check [-top <top>]
prep:
proc
opt_const
opt_clean
check
opt -keepdc
wreduce
memory_dff [-nordff]
opt_clean
memory_collect
opt -keepdc -fast
check:
stat
check
\end{lstlisting}
\section{proc -- translate processes to netlists}
\label{cmd:proc}
\begin{lstlisting}[numbers=left,frame=single]
proc [options] [selection]
This pass calls all the other proc_* passes in the most common order.
proc_clean
proc_rmdead
proc_init
proc_arst
proc_mux
proc_dlatch
proc_dff
proc_clean
This replaces the processes in the design with multiplexers,
flip-flops and latches.
The following options are supported:
-global_arst [!]<netname>
This option is passed through to proc_arst.
\end{lstlisting}
\section{proc\_arst -- detect asynchronous resets}
\label{cmd:proc_arst}
\begin{lstlisting}[numbers=left,frame=single]
proc_arst [-global_arst [!]<netname>] [selection]
This pass identifies asynchronous resets in the processes and converts them
to a different internal representation that is suitable for generating
flip-flop cells with asynchronous resets.
-global_arst [!]<netname>
In modules that have a net with the given name, use this net as async
reset for registers that have been assign initial values in their
declaration ('reg foobar = constant_value;'). Use the '!' modifier for
active low reset signals. Note: the frontend stores the default value
in the 'init' attribute on the net.
\end{lstlisting}
\section{proc\_clean -- remove empty parts of processes}
\label{cmd:proc_clean}
\begin{lstlisting}[numbers=left,frame=single]
proc_clean [selection]
This pass removes empty parts of processes and ultimately removes a process
if it contains only empty structures.
\end{lstlisting}
\section{proc\_dff -- extract flip-flops from processes}
\label{cmd:proc_dff}
\begin{lstlisting}[numbers=left,frame=single]
proc_dff [selection]
This pass identifies flip-flops in the processes and converts them to
d-type flip-flop cells.
\end{lstlisting}
\section{proc\_dlatch -- extract latches from processes}
\label{cmd:proc_dlatch}
\begin{lstlisting}[numbers=left,frame=single]
proc_dlatch [selection]
This pass identifies latches in the processes and converts them to
d-type latches.
\end{lstlisting}
\section{proc\_init -- convert initial block to init attributes}
\label{cmd:proc_init}
\begin{lstlisting}[numbers=left,frame=single]
proc_init [selection]
This pass extracts the 'init' actions from processes (generated from Verilog
'initial' blocks) and sets the initial value to the 'init' attribute on the
respective wire.
\end{lstlisting}
\section{proc\_mux -- convert decision trees to multiplexers}
\label{cmd:proc_mux}
\begin{lstlisting}[numbers=left,frame=single]
proc_mux [selection]
This pass converts the decision trees in processes (originating from if-else
and case statements) to trees of multiplexer cells.
\end{lstlisting}
\section{proc\_rmdead -- eliminate dead trees in decision trees}
\label{cmd:proc_rmdead}
\begin{lstlisting}[numbers=left,frame=single]
proc_rmdead [selection]
This pass identifies unreachable branches in decision trees and removes them.
\end{lstlisting}
\section{qwp -- quadratic wirelength placer}
\label{cmd:qwp}
\begin{lstlisting}[numbers=left,frame=single]
qwp [options] [selection]
This command runs quadratic wirelength placement on the selected modules and
annotates the cells in the design with 'qwp_position' attributes.
-ltr
Add left-to-right constraints: constrain all inputs on the left border
outputs to the right border.
-alpha
Add constraints for inputs/outputs to be placed in alphanumerical
order along the y-axis (top-to-bottom).
-grid N
Number of grid divisions in x- and y-direction. (default=16)
-dump <html_file_name>
Dump a protocol of the placement algorithm to the html file.
Note: This implementation of a quadratic wirelength placer uses exact
dense matrix operations. It is only a toy-placer for small circuits.
\end{lstlisting}
\section{read\_blif -- read BLIF file}
\label{cmd:read_blif}
\begin{lstlisting}[numbers=left,frame=single]
read_blif [filename]
Load modules from a BLIF file into the current design.
\end{lstlisting}
\section{read\_ilang -- read modules from ilang file}
\label{cmd:read_ilang}
\begin{lstlisting}[numbers=left,frame=single]
read_ilang [filename]
Load modules from an ilang file to the current design. (ilang is a text
representation of a design in yosys's internal format.)
\end{lstlisting}
\section{read\_liberty -- read cells from liberty file}
\label{cmd:read_liberty}
\begin{lstlisting}[numbers=left,frame=single]
read_liberty [filename]
Read cells from liberty file as modules into current design.
-lib
only create empty blackbox modules
-ignore_redef
ignore re-definitions of modules. (the default behavior is to
create an error message.)
-ignore_miss_func
ignore cells with missing function specification of outputs
-ignore_miss_dir
ignore cells with a missing or invalid direction
specification on a pin
-setattr <attribute_name>
set the specified attribute (to the value 1) on all loaded modules
\end{lstlisting}
\section{read\_verilog -- read modules from Verilog file}
\label{cmd:read_verilog}
\begin{lstlisting}[numbers=left,frame=single]
read_verilog [options] [filename]
Load modules from a Verilog file to the current design. A large subset of
Verilog-2005 is supported.
-sv
enable support for SystemVerilog features. (only a small subset
of SystemVerilog is supported)
-formal
enable support for assert() and assume() from SystemVerilog
replace the implicit -D SYNTHESIS with -D FORMAL
-dump_ast1
dump abstract syntax tree (before simplification)
-dump_ast2
dump abstract syntax tree (after simplification)
-dump_vlog
dump ast as Verilog code (after simplification)
-yydebug
enable parser debug output
-nolatches
usually latches are synthesized into logic loops
this option prohibits this and sets the output to 'x'
in what would be the latches hold condition
this behavior can also be achieved by setting the
'nolatches' attribute on the respective module or
always block.
-nomem2reg
under certain conditions memories are converted to registers
early during simplification to ensure correct handling of
complex corner cases. this option disables this behavior.
this can also be achieved by setting the 'nomem2reg'
attribute on the respective module or register.
This is potentially dangerous. Usually the front-end has good
reasons for converting an array to a list of registers.
Prohibiting this step will likely result in incorrect synthesis
results.
-mem2reg
always convert memories to registers. this can also be
achieved by setting the 'mem2reg' attribute on the respective
module or register.
-nomeminit
do not infer $meminit cells and instead convert initialized
memories to registers directly in the front-end.
-ppdump
dump Verilog code after pre-processor
-nopp
do not run the pre-processor
-nodpi
disable DPI-C support
-lib
only create empty blackbox modules. This implies -DBLACKBOX.
-noopt
don't perform basic optimizations (such as const folding) in the
high-level front-end.
-icells
interpret cell types starting with '$' as internal cell types
-ignore_redef
ignore re-definitions of modules. (the default behavior is to
create an error message.)
-defer
only read the abstract syntax tree and defer actual compilation
to a later 'hierarchy' command. Useful in cases where the default
parameters of modules yield invalid or not synthesizable code.
-noautowire
make the default of `default_nettype be "none" instead of "wire".
-setattr <attribute_name>
set the specified attribute (to the value 1) on all loaded modules
-Dname[=definition]
define the preprocessor symbol 'name' and set its optional value
'definition'
-Idir
add 'dir' to the directories which are used when searching include
files
The command 'verilog_defaults' can be used to register default options for
subsequent calls to 'read_verilog'.
Note that the Verilog frontend does a pretty good job of processing valid
verilog input, but has not very good error reporting. It generally is
recommended to use a simulator (for example Icarus Verilog) for checking
the syntax of the code, rather than to rely on read_verilog for that.
\end{lstlisting}
\section{rename -- rename object in the design}
\label{cmd:rename}
\begin{lstlisting}[numbers=left,frame=single]
rename old_name new_name
Rename the specified object. Note that selection patterns are not supported
by this command.
rename -enumerate [-pattern <pattern>] [selection]
Assign short auto-generated names to all selected wires and cells with private
names. The -pattern option can be used to set the pattern for the new names.
The character % in the pattern is replaced with a integer number. The default
pattern is '_%_'.
rename -hide [selection]
Assign private names (the ones with $-prefix) to all selected wires and cells
with public names. This ignores all selected ports.
rename -top new_name
Rename top module.
\end{lstlisting}
\section{sat -- solve a SAT problem in the circuit}
\label{cmd:sat}
\begin{lstlisting}[numbers=left,frame=single]
sat [options] [selection]
This command solves a SAT problem defined over the currently selected circuit
and additional constraints passed as parameters.
-all
show all solutions to the problem (this can grow exponentially, use
-max <N> instead to get <N> solutions)
-max <N>
like -all, but limit number of solutions to <N>
-enable_undef
enable modeling of undef value (aka 'x-bits')
this option is implied by -set-def, -set-undef et. cetera
-max_undef
maximize the number of undef bits in solutions, giving a better
picture of which input bits are actually vital to the solution.
-set <signal> <value>
set the specified signal to the specified value.
-set-def <signal>
add a constraint that all bits of the given signal must be defined
-set-any-undef <signal>
add a constraint that at least one bit of the given signal is undefined
-set-all-undef <signal>
add a constraint that all bits of the given signal are undefined
-set-def-inputs
add -set-def constraints for all module inputs
-show <signal>
show the model for the specified signal. if no -show option is
passed then a set of signals to be shown is automatically selected.
-show-inputs, -show-outputs, -show-ports
add all module (input/output) ports to the list of shown signals
-show-regs, -show-public, -show-all
show all registers, show signals with 'public' names, show all signals
-ignore_div_by_zero
ignore all solutions that involve a division by zero
-ignore_unknown_cells
ignore all cells that can not be matched to a SAT model
The following options can be used to set up a sequential problem:
-seq <N>
set up a sequential problem with <N> time steps. The steps will
be numbered from 1 to N.
note: for large <N> it can be significantly faster to use
-tempinduct-baseonly -maxsteps <N> instead of -seq <N>.
-set-at <N> <signal> <value>
-unset-at <N> <signal>
set or unset the specified signal to the specified value in the
given timestep. this has priority over a -set for the same signal.
-set-assumes
set all assumptions provided via $assume cells
-set-def-at <N> <signal>
-set-any-undef-at <N> <signal>
-set-all-undef-at <N> <signal>
add undef constraints in the given timestep.
-set-init <signal> <value>
set the initial value for the register driving the signal to the value
-set-init-undef
set all initial states (not set using -set-init) to undef
-set-init-def
do not force a value for the initial state but do not allow undef
-set-init-zero
set all initial states (not set using -set-init) to zero
-dump_vcd <vcd-file-name>
dump SAT model (counter example in proof) to VCD file
-dump_json <json-file-name>
dump SAT model (counter example in proof) to a WaveJSON file.
-dump_cnf <cnf-file-name>
dump CNF of SAT problem (in DIMACS format). in temporal induction
proofs this is the CNF of the first induction step.
The following additional options can be used to set up a proof. If also -seq
is passed, a temporal induction proof is performed.
-tempinduct
Perform a temporal induction proof. In a temporal induction proof it is
proven that the condition holds forever after the number of time steps
specified using -seq.
-tempinduct-def
Perform a temporal induction proof. Assume an initial state with all
registers set to defined values for the induction step.
-tempinduct-baseonly
Run only the basecase half of temporal induction (requires -maxsteps)
-tempinduct-inductonly
Run only the induction half of temporal induction
-tempinduct-skip <N>
Skip the first <N> steps of the induction proof.
note: this will assume that the base case holds for <N> steps.
this must be proven independently with "-tempinduct-baseonly
-maxsteps <N>". Use -initsteps if you just want to set a
minimal induction length.
-prove <signal> <value>
Attempt to proof that <signal> is always <value>.
-prove-x <signal> <value>
Like -prove, but an undef (x) bit in the lhs matches any value on
the right hand side. Useful for equivalence checking.
-prove-asserts
Prove that all asserts in the design hold.
-prove-skip <N>
Do not enforce the prove-condition for the first <N> time steps.
-maxsteps <N>
Set a maximum length for the induction.
-initsteps <N>
Set initial length for the induction.
This will speed up the search of the right induction length
for deep induction proofs.
-stepsize <N>
Increase the size of the induction proof in steps of <N>.
This will speed up the search of the right induction length
for deep induction proofs.
-timeout <N>
Maximum number of seconds a single SAT instance may take.
-verify
Return an error and stop the synthesis script if the proof fails.
-verify-no-timeout
Like -verify but do not return an error for timeouts.
-falsify
Return an error and stop the synthesis script if the proof succeeds.
-falsify-no-timeout
Like -falsify but do not return an error for timeouts.
\end{lstlisting}
\section{scatter -- add additional intermediate nets}
\label{cmd:scatter}
\begin{lstlisting}[numbers=left,frame=single]
scatter [selection]
This command adds additional intermediate nets on all cell ports. This is used
for testing the correct use of the SigMap helper in passes. If you don't know
what this means: don't worry -- you only need this pass when testing your own
extensions to Yosys.
Use the opt_clean command to get rid of the additional nets.
\end{lstlisting}
\section{scc -- detect strongly connected components (logic loops)}
\label{cmd:scc}
\begin{lstlisting}[numbers=left,frame=single]
scc [options] [selection]
This command identifies strongly connected components (aka logic loops) in the
design.
-expect <num>
expect to find exactly <num> SSCs. A different number of SSCs will
produce an error.
-max_depth <num>
limit to loops not longer than the specified number of cells. This
can e.g. be useful in identifying small local loops in a module that
implements one large SCC.
-nofeedback
do not count cells that have their output fed back into one of their
inputs as single-cell scc.
-all_cell_types
Usually this command only considers internal non-memory cells. With
this option set, all cells are considered. For unknown cells all ports
are assumed to be bidirectional 'inout' ports.
-set_attr <name> <value>
-set_cell_attr <name> <value>
-set_wire_attr <name> <value>
set the specified attribute on all cells and/or wires that are part of
a logic loop. the special token {} in the value is replaced with a
unique identifier for the logic loop.
-select
replace the current selection with a selection of all cells and wires
that are part of a found logic loop
\end{lstlisting}
\section{script -- execute commands from script file}
\label{cmd:script}
\begin{lstlisting}[numbers=left,frame=single]
script <filename> [<from_label>:<to_label>]
This command executes the yosys commands in the specified file.
The 2nd argument can be used to only execute the section of the
file between the specified labels. An empty from label is synonymous
for the beginning of the file and an empty to label is synonymous
for the end of the file.
If only one label is specified (without ':') then only the block
marked with that label (until the next label) is executed.
\end{lstlisting}
\section{select -- modify and view the list of selected objects}
\label{cmd:select}
\begin{lstlisting}[numbers=left,frame=single]
select [ -add | -del | -set <name> ] {-read <filename> | <selection>}
select [ -assert-none | -assert-any ] {-read <filename> | <selection>}
select [ -list | -write <filename> | -count | -clear ]
select -module <modname>
Most commands use the list of currently selected objects to determine which part
of the design to operate on. This command can be used to modify and view this
list of selected objects.
Note that many commands support an optional [selection] argument that can be
used to override the global selection for the command. The syntax of this
optional argument is identical to the syntax of the <selection> argument
described here.
-add, -del
add or remove the given objects to the current selection.
without this options the current selection is replaced.
-set <name>
do not modify the current selection. instead save the new selection
under the given name (see @<name> below). to save the current selection,
use "select -set <name> %"
-assert-none
do not modify the current selection. instead assert that the given
selection is empty. i.e. produce an error if any object matching the
selection is found.
-assert-any
do not modify the current selection. instead assert that the given
selection is non-empty. i.e. produce an error if no object matching
the selection is found.
-assert-count N
do not modify the current selection. instead assert that the given
selection contains exactly N objects.
-list
list all objects in the current selection
-write <filename>
like -list but write the output to the specified file
-read <filename>
read the specified file (written by -write)
-count
count all objects in the current selection
-clear
clear the current selection. this effectively selects the whole
design. it also resets the selected module (see -module). use the
command 'select *' to select everything but stay in the current module.
-none
create an empty selection. the current module is unchanged.
-module <modname>
limit the current scope to the specified module.
the difference between this and simply selecting the module
is that all object names are interpreted relative to this
module after this command until the selection is cleared again.
When this command is called without an argument, the current selection
is displayed in a compact form (i.e. only the module name when a whole module
is selected).
The <selection> argument itself is a series of commands for a simple stack
machine. Each element on the stack represents a set of selected objects.
After this commands have been executed, the union of all remaining sets
on the stack is computed and used as selection for the command.
Pushing (selecting) object when not in -module mode:
<mod_pattern>
select the specified module(s)
<mod_pattern>/<obj_pattern>
select the specified object(s) from the module(s)
Pushing (selecting) object when in -module mode:
<obj_pattern>
select the specified object(s) from the current module
A <mod_pattern> can be a module name, wildcard expression (*, ?, [..])
matching module names, or one of the following:
A:<pattern>, A:<pattern>=<pattern>
all modules with an attribute matching the given pattern
in addition to = also <, <=, >=, and > are supported
An <obj_pattern> can be an object name, wildcard expression, or one of
the following:
w:<pattern>
all wires with a name matching the given wildcard pattern
i:<pattern>, o:<pattern>, x:<pattern>
all inputs (i:), outputs (o:) or any ports (x:) with matching names
s:<size>, s:<min>:<max>
all wires with a matching width
m:<pattern>
all memories with a name matching the given pattern
c:<pattern>
all cells with a name matching the given pattern
t:<pattern>
all cells with a type matching the given pattern
p:<pattern>
all processes with a name matching the given pattern
a:<pattern>
all objects with an attribute name matching the given pattern
a:<pattern>=<pattern>
all objects with a matching attribute name-value-pair.
in addition to = also <, <=, >=, and > are supported
r:<pattern>, r:<pattern>=<pattern>
cells with matching parameters. also with <, <=, >= and >.
n:<pattern>
all objects with a name matching the given pattern
(i.e. 'n:' is optional as it is the default matching rule)
@<name>
push the selection saved prior with 'select -set <name> ...'
The following actions can be performed on the top sets on the stack:
%
push a copy of the current selection to the stack
%%
replace the stack with a union of all elements on it
%n
replace top set with its invert
%u
replace the two top sets on the stack with their union
%i
replace the two top sets on the stack with their intersection
%d
pop the top set from the stack and subtract it from the new top
%D
like %d but swap the roles of two top sets on the stack
%c
create a copy of the top set from the stack and push it
%x[<num1>|*][.<num2>][:<rule>[:<rule>..]]
expand top set <num1> num times according to the specified rules.
(i.e. select all cells connected to selected wires and select all
wires connected to selected cells) The rules specify which cell
ports to use for this. the syntax for a rule is a '-' for exclusion
and a '+' for inclusion, followed by an optional comma separated
list of cell types followed by an optional comma separated list of
cell ports in square brackets. a rule can also be just a cell or wire
name that limits the expansion (is included but does not go beyond).
select at most <num2> objects. a warning message is printed when this
limit is reached. When '*' is used instead of <num1> then the process
is repeated until no further object are selected.
%ci[<num1>|*][.<num2>][:<rule>[:<rule>..]]
%co[<num1>|*][.<num2>][:<rule>[:<rule>..]]
similar to %x, but only select input (%ci) or output cones (%co)
%xe[...] %cie[...] %coe
like %x, %ci, and %co but only consider combinatorial cells
%a
expand top set by selecting all wires that are (at least in part)
aliases for selected wires.
%s
expand top set by adding all modules that implement cells in selected
modules
%m
expand top set by selecting all modules that contain selected objects
%M
select modules that implement selected cells
%C
select cells that implement selected modules
%R[<num>]
select <num> random objects from top selection (default 1)
Example: the following command selects all wires that are connected to a
'GATE' input of a 'SWITCH' cell:
select */t:SWITCH %x:+[GATE] */t:SWITCH %d
\end{lstlisting}
\section{setattr -- set/unset attributes on objects}
\label{cmd:setattr}
\begin{lstlisting}[numbers=left,frame=single]
setattr [ -mod ] [ -set name value | -unset name ]... [selection]
Set/unset the given attributes on the selected objects. String values must be
passed in double quotes (").
When called with -mod, this command will set and unset attributes on modules
instead of objects within modules.
\end{lstlisting}
\section{setparam -- set/unset parameters on objects}
\label{cmd:setparam}
\begin{lstlisting}[numbers=left,frame=single]
setparam [ -set name value | -unset name ]... [selection]
Set/unset the given parameters on the selected cells. String values must be
passed in double quotes (").
\end{lstlisting}
\section{setundef -- replace undef values with defined constants}
\label{cmd:setundef}
\begin{lstlisting}[numbers=left,frame=single]
setundef [options] [selection]
This command replaced undef (x) constants with defined (0/1) constants.
-undriven
also set undriven nets to constant values
-zero
replace with bits cleared (0)
-one
replace with bits set (1)
-random <seed>
replace with random bits using the specified integer als seed
value for the random number generator.
\end{lstlisting}
\section{share -- perform sat-based resource sharing}
\label{cmd:share}
\begin{lstlisting}[numbers=left,frame=single]
share [options] [selection]
This pass merges shareable resources into a single resource. A SAT solver
is used to determine if two resources are share-able.
-force
Per default the selection of cells that is considered for sharing is
narrowed using a list of cell types. With this option all selected
cells are considered for resource sharing.
IMPORTANT NOTE: If the -all option is used then no cells with internal
state must be selected!
-aggressive
Per default some heuristics are used to reduce the number of cells
considered for resource sharing to only large resources. This options
turns this heuristics off, resulting in much more cells being considered
for resource sharing.
-fast
Only consider the simple part of the control logic in SAT solving, resulting
in much easier SAT problems at the cost of maybe missing some opportunities
for resource sharing.
-limit N
Only perform the first N merges, then stop. This is useful for debugging.
\end{lstlisting}
\section{shell -- enter interactive command mode}
\label{cmd:shell}
\begin{lstlisting}[numbers=left,frame=single]
shell
This command enters the interactive command mode. This can be useful
in a script to interrupt the script at a certain point and allow for
interactive inspection or manual synthesis of the design at this point.
The command prompt of the interactive shell indicates the current
selection (see 'help select'):
yosys>
the entire design is selected
yosys*>
only part of the design is selected
yosys [modname]>
the entire module 'modname' is selected using 'select -module modname'
yosys [modname]*>
only part of current module 'modname' is selected
When in interactive shell, some errors (e.g. invalid command arguments)
do not terminate yosys but return to the command prompt.
This command is the default action if nothing else has been specified
on the command line.
Press Ctrl-D or type 'exit' to leave the interactive shell.
\end{lstlisting}
\section{show -- generate schematics using graphviz}
\label{cmd:show}
\begin{lstlisting}[numbers=left,frame=single]
show [options] [selection]
Create a graphviz DOT file for the selected part of the design and compile it
to a graphics file (usually SVG or PostScript).
-viewer <viewer>
Run the specified command with the graphics file as parameter.
-format <format>
Generate a graphics file in the specified format.
Usually <format> is 'svg' or 'ps'.
-lib <verilog_or_ilang_file>
Use the specified library file for determining whether cell ports are
inputs or outputs. This option can be used multiple times to specify
more than one library.
note: in most cases it is better to load the library before calling
show with 'read_verilog -lib <filename>'. it is also possible to
load liberty files with 'read_liberty -lib <filename>'.
-prefix <prefix>
generate <prefix>.* instead of ~/.yosys_show.*
-color <color> <object>
assign the specified color to the specified object. The object can be
a single selection wildcard expressions or a saved set of objects in
the @<name> syntax (see "help select" for details).
-label <text> <object>
assign the specified label text to the specified object. The object can
be a single selection wildcard expressions or a saved set of objects in
the @<name> syntax (see "help select" for details).
-colors <seed>
Randomly assign colors to the wires. The integer argument is the seed
for the random number generator. Change the seed value if the colored
graph still is ambiguous. A seed of zero deactivates the coloring.
-colorattr <attribute_name>
Use the specified attribute to assign colors. A unique color is
assigned to each unique value of this attribute.
-width
annotate busses with a label indicating the width of the bus.
-signed
mark ports (A, B) that are declared as signed (using the [AB]_SIGNED
cell parameter) with an asterisk next to the port name.
-stretch
stretch the graph so all inputs are on the left side and all outputs
(including inout ports) are on the right side.
-pause
wait for the use to press enter to before returning
-enum
enumerate objects with internal ($-prefixed) names
-long
do not abbreviate objects with internal ($-prefixed) names
-notitle
do not add the module name as graph title to the dot file
When no <format> is specified, 'dot' is used. When no <format> and <viewer> is
specified, 'xdot' is used to display the schematic.
The generated output files are '~/.yosys_show.dot' and '~/.yosys_show.<format>',
unless another prefix is specified using -prefix <prefix>.
Yosys on Windows and YosysJS use different defaults: The output is written
to 'show.dot' in the current directory and new viewer is launched.
\end{lstlisting}
\section{simplemap -- mapping simple coarse-grain cells}
\label{cmd:simplemap}
\begin{lstlisting}[numbers=left,frame=single]
simplemap [selection]
This pass maps a small selection of simple coarse-grain cells to yosys gate
primitives. The following internal cell types are mapped by this pass:
$not, $pos, $and, $or, $xor, $xnor
$reduce_and, $reduce_or, $reduce_xor, $reduce_xnor, $reduce_bool
$logic_not, $logic_and, $logic_or, $mux, $tribuf
$sr, $dff, $dffsr, $adff, $dlatch
\end{lstlisting}
\section{singleton -- create singleton modules}
\label{cmd:singleton}
\begin{lstlisting}[numbers=left,frame=single]
singleton [selection]
By default, a module that is instantiated by several other modules is only
kept once in the design. This preserves the original modularity of the design
and reduces the overall size of the design in memory. But it prevents certain
optimizations and other operations on the design. This pass creates singleton
modules for all selected cells. The created modules are marked with the
'singleton' attribute.
This commands only operates on modules that by themself have the 'singleton'
attribute set (the 'top' module is a singleton implicitly).
\end{lstlisting}
\section{splice -- create explicit splicing cells}
\label{cmd:splice}
\begin{lstlisting}[numbers=left,frame=single]
splice [options] [selection]
This command adds $slice and $concat cells to the design to make the splicing
of multi-bit signals explicit. This for example is useful for coarse grain
synthesis, where dedicated hardware is needed to splice signals.
-sel_by_cell
only select the cell ports to rewire by the cell. if the selection
contains a cell, than all cell inputs are rewired, if necessary.
-sel_by_wire
only select the cell ports to rewire by the wire. if the selection
contains a wire, than all cell ports driven by this wire are wired,
if necessary.
-sel_any_bit
it is sufficient if the driver of any bit of a cell port is selected.
by default all bits must be selected.
-wires
also add $slice and $concat cells to drive otherwise unused wires.
-no_outputs
do not rewire selected module outputs.
-port <name>
only rewire cell ports with the specified name. can be used multiple
times. implies -no_output.
-no_port <name>
do not rewire cell ports with the specified name. can be used multiple
times. can not be combined with -port <name>.
By default selected output wires and all cell ports of selected cells driven
by selected wires are rewired.
\end{lstlisting}
\section{splitnets -- split up multi-bit nets}
\label{cmd:splitnets}
\begin{lstlisting}[numbers=left,frame=single]
splitnets [options] [selection]
This command splits multi-bit nets into single-bit nets.
-format char1[char2[char3]]
the first char is inserted between the net name and the bit index, the
second char is appended to the netname. e.g. -format () creates net
names like 'mysignal(42)'. the 3rd character is the range separation
character when creating multi-bit wires. the default is '[]:'.
-ports
also split module ports. per default only internal signals are split.
-driver
don't blindly split nets in individual bits. instead look at the driver
and split nets so that no driver drives only part of a net.
\end{lstlisting}
\section{stat -- print some statistics}
\label{cmd:stat}
\begin{lstlisting}[numbers=left,frame=single]
stat [options] [selection]
Print some statistics (number of objects) on the selected portion of the
design.
-top <module>
print design hierarchy with this module as top. if the design is fully
selected and a module has the 'top' attribute set, this module is used
default value for this option.
-liberty <liberty_file>
use cell area information from the provided liberty file
-width
annotate internal cell types with their word width.
e.g. $add_8 for an 8 bit wide $add cell.
\end{lstlisting}
\section{submod -- moving part of a module to a new submodule}
\label{cmd:submod}
\begin{lstlisting}[numbers=left,frame=single]
submod [-copy] [selection]
This pass identifies all cells with the 'submod' attribute and moves them to
a newly created module. The value of the attribute is used as name for the
cell that replaces the group of cells with the same attribute value.
This pass can be used to create a design hierarchy in flat design. This can
be useful for analyzing or reverse-engineering a design.
This pass only operates on completely selected modules with no processes
or memories.
submod -name <name> [-copy] [selection]
As above, but don't use the 'submod' attribute but instead use the selection.
Only objects from one module might be selected. The value of the -name option
is used as the value of the 'submod' attribute above.
By default the cells are 'moved' from the source module and the source module
will use an instance of the new module after this command is finished. Call
with -copy to not modify the source module.
\end{lstlisting}
\section{synth -- generic synthesis script}
\label{cmd:synth}
\begin{lstlisting}[numbers=left,frame=single]
synth [options]
This command runs the default synthesis script. This command does not operate
on partly selected designs.
-top <module>
use the specified module as top module (default='top')
-encfile <file>
passed to 'fsm_recode' via 'fsm'
-nofsm
do not run FSM optimization
-noabc
do not run abc (as if yosys was compiled without ABC support)
-noalumacc
do not run 'alumacc' pass. i.e. keep arithmetic operators in
their direct form ($add, $sub, etc.).
-nordff
passed to 'memory'. prohibits merging of FFs into memory read ports
-run <from_label>[:<to_label>]
only run the commands between the labels (see below). an empty
from label is synonymous to 'begin', and empty to label is
synonymous to the end of the command list.
The following commands are executed by this synthesis command:
begin:
hierarchy -check [-top <top>]
coarse:
proc
opt_const
opt_clean
check
opt
wreduce
alumacc
share
opt
fsm
opt -fast
memory -nomap
opt_clean
fine:
opt -fast -full
memory_map
opt -full
techmap
opt -fast
abc -fast
opt -fast
check:
hierarchy -check
stat
check
\end{lstlisting}
\section{synth\_greenpak4 -- synthesis for GreenPAK4 FPGAs}
\label{cmd:synth_greenpak4}
\begin{lstlisting}[numbers=left,frame=single]
synth_greenpak4 [options]
This command runs synthesis for GreenPAK4 FPGAs. This work is experimental.
-top <module>
use the specified module as top module (default='top')
-blif <file>
write the design to the specified BLIF file. writing of an output file
is omitted if this parameter is not specified.
-edif <file>
write the design to the specified edif file. writing of an output file
is omitted if this parameter is not specified.
-run <from_label>:<to_label>
only run the commands between the labels (see below). an empty
from label is synonymous to 'begin', and empty to label is
synonymous to the end of the command list.
-noflatten
do not flatten design before synthesis
-retime
run 'abc' with -dff option
The following commands are executed by this synthesis command:
begin:
read_verilog -lib +/greenpak4/cells_sim.v
hierarchy -check -top <top>
flatten: (unless -noflatten)
proc
flatten
tribuf -logic
coarse:
synth -run coarse
fine:
opt -fast -mux_undef -undriven -fine
memory_map
opt -undriven -fine
techmap
dfflibmap -prepare -liberty +/greenpak4/gp_dff.lib
opt -fast
abc -dff (only if -retime)
map_luts:
nlutmap -luts 0,8,16,2
clean
map_cells:
techmap -map +/greenpak4/cells_map.v
clean
check:
hierarchy -check
stat
check -noinit
blif:
write_blif -gates -attr -param <file-name>
edif:
write_edif <file-name>
\end{lstlisting}
\section{synth\_ice40 -- synthesis for iCE40 FPGAs}
\label{cmd:synth_ice40}
\begin{lstlisting}[numbers=left,frame=single]
synth_ice40 [options]
This command runs synthesis for iCE40 FPGAs. This work is experimental.
-top <module>
use the specified module as top module (default='top')
-blif <file>
write the design to the specified BLIF file. writing of an output file
is omitted if this parameter is not specified.
-edif <file>
write the design to the specified edif file. writing of an output file
is omitted if this parameter is not specified.
-run <from_label>:<to_label>
only run the commands between the labels (see below). an empty
from label is synonymous to 'begin', and empty to label is
synonymous to the end of the command list.
-noflatten
do not flatten design before synthesis
-retime
run 'abc' with -dff option
-nocarry
do not use SB_CARRY cells in output netlist
-nobram
do not use SB_RAM40_4K* cells in output netlist
-abc2
run two passes of 'abc' for slightly improved logic density
The following commands are executed by this synthesis command:
begin:
read_verilog -lib +/ice40/cells_sim.v
hierarchy -check -top <top>
flatten: (unless -noflatten)
proc
flatten
tribuf -logic
coarse:
synth -run coarse
bram: (skip if -nobram)
memory_bram -rules +/ice40/brams.txt
techmap -map +/ice40/brams_map.v
fine:
opt -fast -mux_undef -undriven -fine
memory_map
opt -undriven -fine
techmap -map +/techmap.v [-map +/ice40/arith_map.v]
abc -dff (only if -retime)
ice40_opt
map_ffs:
dffsr2dff
dff2dffe -direct-match $_DFF_*
techmap -map +/ice40/cells_map.v
opt_const -mux_undef
simplemap
ice40_ffinit
ice40_ffssr
ice40_opt -full
map_luts:
abc (only if -abc2)
ice40_opt (only if -abc2)
abc -lut 4
clean
map_cells:
techmap -map +/ice40/cells_map.v
clean
check:
hierarchy -check
stat
check -noinit
blif:
write_blif -gates -attr -param <file-name>
edif:
write_edif <file-name>
\end{lstlisting}
\section{synth\_xilinx -- synthesis for Xilinx FPGAs}
\label{cmd:synth_xilinx}
\begin{lstlisting}[numbers=left,frame=single]
synth_xilinx [options]
This command runs synthesis for Xilinx FPGAs. This command does not operate on
partly selected designs. At the moment this command creates netlists that are
compatible with 7-Series Xilinx devices.
-top <module>
use the specified module as top module
-edif <file>
write the design to the specified edif file. writing of an output file
is omitted if this parameter is not specified.
-run <from_label>:<to_label>
only run the commands between the labels (see below). an empty
from label is synonymous to 'begin', and empty to label is
synonymous to the end of the command list.
-flatten
flatten design before synthesis
-retime
run 'abc' with -dff option
The following commands are executed by this synthesis command:
begin:
read_verilog -lib +/xilinx/cells_sim.v
read_verilog -lib +/xilinx/brams_bb.v
read_verilog -lib +/xilinx/drams_bb.v
hierarchy -check -top <top>
flatten: (only if -flatten)
proc
flatten
coarse:
synth -run coarse
bram:
memory_bram -rules +/xilinx/brams.txt
techmap -map +/xilinx/brams_map.v
dram:
memory_bram -rules +/xilinx/drams.txt
techmap -map +/xilinx/drams_map.v
fine:
opt -fast -full
memory_map
dffsr2dff
dff2dffe
opt -full
techmap -map +/techmap.v -map +/xilinx/arith_map.v
opt -fast
map_luts:
abc -luts 2:2,3,6:5,10,20 [-dff]
clean
map_cells:
techmap -map +/xilinx/cells_map.v
dffinit -ff FDRE Q INIT -ff FDCE Q INIT -ff FDPE Q INIT
clean
check:
hierarchy -check
stat
check -noinit
edif: (only if -edif)
write_edif <file-name>
\end{lstlisting}
\section{tcl -- execute a TCL script file}
\label{cmd:tcl}
\begin{lstlisting}[numbers=left,frame=single]
tcl <filename>
This command executes the tcl commands in the specified file.
Use 'yosys cmd' to run the yosys command 'cmd' from tcl.
The tcl command 'yosys -import' can be used to import all yosys
commands directly as tcl commands to the tcl shell. The yosys
command 'proc' is wrapped using the tcl command 'procs' in order
to avoid a name collision with the tcl builtin command 'proc'.
\end{lstlisting}
\section{techmap -- generic technology mapper}
\label{cmd:techmap}
\begin{lstlisting}[numbers=left,frame=single]
techmap [-map filename] [selection]
This pass implements a very simple technology mapper that replaces cells in
the design with implementations given in form of a Verilog or ilang source
file.
-map filename
the library of cell implementations to be used.
without this parameter a builtin library is used that
transforms the internal RTL cells to the internal gate
library.
-map %<design-name>
like -map above, but with an in-memory design instead of a file.
-extern
load the cell implementations as separate modules into the design
instead of inlining them.
-max_iter <number>
only run the specified number of iterations.
-recursive
instead of the iterative breadth-first algorithm use a recursive
depth-first algorithm. both methods should yield equivalent results,
but may differ in performance.
-autoproc
Automatically call "proc" on implementations that contain processes.
-assert
this option will cause techmap to exit with an error if it can't map
a selected cell. only cell types that end on an underscore are accepted
as final cell types by this mode.
-D <define>, -I <incdir>
this options are passed as-is to the Verilog frontend for loading the
map file. Note that the Verilog frontend is also called with the
'-ignore_redef' option set.
When a module in the map file has the 'techmap_celltype' attribute set, it will
match cells with a type that match the text value of this attribute. Otherwise
the module name will be used to match the cell.
When a module in the map file has the 'techmap_simplemap' attribute set, techmap
will use 'simplemap' (see 'help simplemap') to map cells matching the module.
When a module in the map file has the 'techmap_maccmap' attribute set, techmap
will use 'maccmap' (see 'help maccmap') to map cells matching the module.
When a module in the map file has the 'techmap_wrap' attribute set, techmap
will create a wrapper for the cell and then run the command string that the
attribute is set to on the wrapper module.
All wires in the modules from the map file matching the pattern _TECHMAP_*
or *._TECHMAP_* are special wires that are used to pass instructions from
the mapping module to the techmap command. At the moment the following special
wires are supported:
_TECHMAP_FAIL_
When this wire is set to a non-zero constant value, techmap will not
use this module and instead try the next module with a matching
'techmap_celltype' attribute.
When such a wire exists but does not have a constant value after all
_TECHMAP_DO_* commands have been executed, an error is generated.
_TECHMAP_DO_*
This wires are evaluated in alphabetical order. The constant text value
of this wire is a yosys command (or sequence of commands) that is run
by techmap on the module. A common use case is to run 'proc' on modules
that are written using always-statements.
When such a wire has a non-constant value at the time it is to be
evaluated, an error is produced. That means it is possible for such a
wire to start out as non-constant and evaluate to a constant value
during processing of other _TECHMAP_DO_* commands.
A _TECHMAP_DO_* command may start with the special token 'CONSTMAP; '.
in this case techmap will create a copy for each distinct configuration
of constant inputs and shorted inputs at this point and import the
constant and connected bits into the map module. All further commands
are executed in this copy. This is a very convenient way of creating
optimized specializations of techmap modules without using the special
parameters described below.
A _TECHMAP_DO_* command may start with the special token 'RECURSION; '.
then techmap will recursively replace the cells in the module with their
implementation. This is not affected by the -max_iter option.
It is possible to combine both prefixes to 'RECURSION; CONSTMAP; '.
In addition to this special wires, techmap also supports special parameters in
modules in the map file:
_TECHMAP_CELLTYPE_
When a parameter with this name exists, it will be set to the type name
of the cell that matches the module.
_TECHMAP_CONSTMSK_<port-name>_
_TECHMAP_CONSTVAL_<port-name>_
When this pair of parameters is available in a module for a port, then
former has a 1-bit for each constant input bit and the latter has the
value for this bit. The unused bits of the latter are set to undef (x).
_TECHMAP_BITS_CONNMAP_
_TECHMAP_CONNMAP_<port-name>_
For an N-bit port, the _TECHMAP_CONNMAP_<port-name>_ parameter, if it
exists, will be set to an N*_TECHMAP_BITS_CONNMAP_ bit vector containing
N words (of _TECHMAP_BITS_CONNMAP_ bits each) that assign each single
bit driver a unique id. The values 0-3 are reserved for 0, 1, x, and z.
This can be used to detect shorted inputs.
When a module in the map file has a parameter where the according cell in the
design has a port, the module from the map file is only used if the port in
the design is connected to a constant value. The parameter is then set to the
constant value.
A cell with the name _TECHMAP_REPLACE_ in the map file will inherit the name
of the cell that is being replaced.
See 'help extract' for a pass that does the opposite thing.
See 'help flatten' for a pass that does flatten the design (which is
essentially techmap but using the design itself as map library).
\end{lstlisting}
\section{tee -- redirect command output to file}
\label{cmd:tee}
\begin{lstlisting}[numbers=left,frame=single]
tee [-q] [-o logfile|-a logfile] cmd
Execute the specified command, optionally writing the commands output to the
specified logfile(s).
-q
Do not print output to the normal destination (console and/or log file)
-o logfile
Write output to this file, truncate if exists.
-a logfile
Write output to this file, append if exists.
\end{lstlisting}
\section{test\_abcloop -- automatically test handling of loops in abc command}
\label{cmd:test_abcloop}
\begin{lstlisting}[numbers=left,frame=single]
test_abcloop [options]
Test handling of logic loops in ABC.
-n {integer}
create this number of circuits and test them (default = 100).
-s {positive_integer}
use this value as rng seed value (default = unix time).
\end{lstlisting}
\section{test\_autotb -- generate simple test benches}
\label{cmd:test_autotb}
\begin{lstlisting}[numbers=left,frame=single]
test_autotb [options] [filename]
Automatically create primitive Verilog test benches for all modules in the
design. The generated testbenches toggle the input pins of the module in
a semi-random manner and dumps the resulting output signals.
This can be used to check the synthesis results for simple circuits by
comparing the testbench output for the input files and the synthesis results.
The backend automatically detects clock signals. Additionally a signal can
be forced to be interpreted as clock signal by setting the attribute
'gentb_clock' on the signal.
The attribute 'gentb_constant' can be used to force a signal to a constant
value after initialization. This can e.g. be used to force a reset signal
low in order to explore more inner states in a state machine.
-n <int>
number of iterations the test bench should run (default = 1000)
\end{lstlisting}
\section{test\_cell -- automatically test the implementation of a cell type}
\label{cmd:test_cell}
\begin{lstlisting}[numbers=left,frame=single]
test_cell [options] {cell-types}
Tests the internal implementation of the given cell type (for example '$add')
by comparing SAT solver, EVAL and TECHMAP implementations of the cell types..
Run with 'all' instead of a cell type to run the test on all supported
cell types. Use for example 'all /$add' for all cell types except $add.
-n {integer}
create this number of cell instances and test them (default = 100).
-s {positive_integer}
use this value as rng seed value (default = unix time).
-f {ilang_file}
don't generate circuits. instead load the specified ilang file.
-w {filename_prefix}
don't test anything. just generate the circuits and write them
to ilang files with the specified prefix
-map {filename}
pass this option to techmap.
-simlib
use "techmap -map +/simlib.v -max_iter 2 -autoproc"
-aigmap
instead of calling "techmap", call "aigmap"
-muxdiv
when creating test benches with dividers, create an additional mux
to mask out the division-by-zero case
-script {script_file}
instead of calling "techmap", call "script {script_file}".
-const
set some input bits to random constant values
-nosat
do not check SAT model or run SAT equivalence checking
-noeval
do not check const-eval models
-v
print additional debug information to the console
-vlog {filename}
create a Verilog test bench to test simlib and write_verilog
\end{lstlisting}
\section{torder -- print cells in topological order}
\label{cmd:torder}
\begin{lstlisting}[numbers=left,frame=single]
torder [options] [selection]
This command prints the selected cells in topological order.
-stop <cell_type> <cell_port>
do not use the specified cell port in topological sorting
-noautostop
by default Q outputs of internal FF cells and memory read port outputs
are not used in topological sorting. this option deactivates that.
\end{lstlisting}
\section{trace -- redirect command output to file}
\label{cmd:trace}
\begin{lstlisting}[numbers=left,frame=single]
trace cmd
Execute the specified command, logging all changes the command performs on
the design in real time.
\end{lstlisting}
\section{tribuf -- infer tri-state buffers}
\label{cmd:tribuf}
\begin{lstlisting}[numbers=left,frame=single]
tribuf [options] [selection]
This pass transforms $mux cells with 'z' inputs to tristate buffers.
-merge
merge multiple tri-state buffers driving the same net
into a single buffer.
-logic
convert tri-state buffers that do not drive output ports
to non-tristate logic. this option implies -merge.
\end{lstlisting}
\section{verific -- load Verilog and VHDL designs using Verific}
\label{cmd:verific}
\begin{lstlisting}[numbers=left,frame=single]
verific {-vlog95|-vlog2k|-sv2005|-sv2009|-sv} <verilog-file>..
Load the specified Verilog/SystemVerilog files into Verific.
verific {-vhdl87|-vhdl93|-vhdl2k|-vhdl2008} <vhdl-file>..
Load the specified VHDL files into Verific.
verific -import [-gates] {-all | <top-module>..}
Elaborate the design for the specified top modules, import to Yosys and
reset the internal state of Verific. A gate-level netlist is created
when called with -gates.
Visit http://verific.com/ for more information on Verific.
\end{lstlisting}
\section{verilog\_defaults -- set default options for read\_verilog}
\label{cmd:verilog_defaults}
\begin{lstlisting}[numbers=left,frame=single]
verilog_defaults -add [options]
Add the specified options to the list of default options to read_verilog.
verilog_defaults -clear
Clear the list of Verilog default options.
verilog_defaults -push verilog_defaults -pop
Push or pop the list of default options to a stack. Note that -push does
not imply -clear.
\end{lstlisting}
\section{vhdl2verilog -- importing VHDL designs using vhdl2verilog}
\label{cmd:vhdl2verilog}
\begin{lstlisting}[numbers=left,frame=single]
vhdl2verilog [options] <vhdl-file>..
This command reads VHDL source files using the 'vhdl2verilog' tool and the
Yosys Verilog frontend.
-out <out_file>
do not import the vhdl2verilog output. instead write it to the
specified file.
-vhdl2verilog_dir <directory>
do use the specified vhdl2verilog installation. this is the directory
that contains the setup_env.sh file. when this option is not present,
it is assumed that vhdl2verilog is in the PATH environment variable.
-top <top-entity-name>
The name of the top entity. This option is mandatory.
The following options are passed as-is to vhdl2verilog:
-arch <architecture_name>
-unroll_generate
-nogenericeval
-nouniquify
-oldparser
-suppress <list>
-quiet
-nobanner
-mapfile <file>
vhdl2verilog can be obtained from:
http://www.edautils.com/vhdl2verilog.html
\end{lstlisting}
\section{wreduce -- reduce the word size of operations if possible}
\label{cmd:wreduce}
\begin{lstlisting}[numbers=left,frame=single]
wreduce [options] [selection]
This command reduces the word size of operations. For example it will replace
the 32 bit adders in the following code with adders of more appropriate widths:
module test(input [3:0] a, b, c, output [7:0] y);
assign y = a + b + c + 1;
endmodule
\end{lstlisting}
\section{write\_blif -- write design to BLIF file}
\label{cmd:write_blif}
\begin{lstlisting}[numbers=left,frame=single]
write_blif [options] [filename]
Write the current design to an BLIF file.
-top top_module
set the specified module as design top module
-buf <cell-type> <in-port> <out-port>
use cells of type <cell-type> with the specified port names for buffers
-unbuf <cell-type> <in-port> <out-port>
replace buffer cells with the specified name and port names with
a .names statement that models a buffer
-true <cell-type> <out-port>
-false <cell-type> <out-port>
-undef <cell-type> <out-port>
use the specified cell types to drive nets that are constant 1, 0, or
undefined. when '-' is used as <cell-type>, then <out-port> specifies
the wire name to be used for the constant signal and no cell driving
that wire is generated.
The following options can be useful when the generated file is not going to be
read by a BLIF parser but a custom tool. It is recommended to not name the output
file *.blif when any of this options is used.
-icells
do not translate Yosys's internal gates to generic BLIF logic
functions. Instead create .subckt or .gate lines for all cells.
-gates
print .gate instead of .subckt lines for all cells that are not
instantiations of other modules from this design.
-conn
do not generate buffers for connected wires. instead use the
non-standard .conn statement.
-attr
use the non-standard .attr statement to write cell attributes
-param
use the non-standard .param statement to write cell parameters
-cname
use the non-standard .cname statement to write cell names
-blackbox
write blackbox cells with .blackbox statement.
-impltf
do not write definitions for the $true, $false and $undef wires.
\end{lstlisting}
\section{write\_btor -- write design to BTOR file}
\label{cmd:write_btor}
\begin{lstlisting}[numbers=left,frame=single]
write_btor [filename]
Write the current design to an BTOR file.
\end{lstlisting}
\section{write\_edif -- write design to EDIF netlist file}
\label{cmd:write_edif}
\begin{lstlisting}[numbers=left,frame=single]
write_edif [options] [filename]
Write the current design to an EDIF netlist file.
-top top_module
set the specified module as design top module
Unfortunately there are different "flavors" of the EDIF file format. This
command generates EDIF files for the Xilinx place&route tools. It might be
necessary to make small modifications to this command when a different tool
is targeted.
\end{lstlisting}
\section{write\_file -- write a text to a file}
\label{cmd:write_file}
\begin{lstlisting}[numbers=left,frame=single]
write_file [options] output_file [input_file]
Write the text from the input file to the output file.
-a
Append to output file (instead of overwriting)
Inside a script the input file can also can a here-document:
write_file hello.txt <<EOT
Hello World!
EOT
\end{lstlisting}
\section{write\_ilang -- write design to ilang file}
\label{cmd:write_ilang}
\begin{lstlisting}[numbers=left,frame=single]
write_ilang [filename]
Write the current design to an 'ilang' file. (ilang is a text representation
of a design in yosys's internal format.)
-selected
only write selected parts of the design.
\end{lstlisting}
\section{write\_intersynth -- write design to InterSynth netlist file}
\label{cmd:write_intersynth}
\begin{lstlisting}[numbers=left,frame=single]
write_intersynth [options] [filename]
Write the current design to an 'intersynth' netlist file. InterSynth is
a tool for Coarse-Grain Example-Driven Interconnect Synthesis.
-notypes
do not generate celltypes and conntypes commands. i.e. just output
the netlists. this is used for postsilicon synthesis.
-lib <verilog_or_ilang_file>
Use the specified library file for determining whether cell ports are
inputs or outputs. This option can be used multiple times to specify
more than one library.
-selected
only write selected modules. modules must be selected entirely or
not at all.
http://www.clifford.at/intersynth/
\end{lstlisting}
\section{write\_json -- write design to a JSON file}
\label{cmd:write_json}
\begin{lstlisting}[numbers=left,frame=single]
write_json [options] [filename]
Write a JSON netlist of the current design.
-aig
include AIG models for the different gate types
The general syntax of the JSON output created by this command is as follows:
{
"modules": {
<module_name>: {
"ports": {
<port_name>: <port_details>,
...
},
"cells": {
<cell_name>: <cell_details>,
...
},
"netnames": {
<net_name>: <net_details>,
...
}
}
},
"models": {
...
},
}
Where <port_details> is:
{
"direction": <"input" | "output" | "inout">,
"bits": <bit_vector>
}
And <cell_details> is:
{
"hide_name": <1 | 0>,
"type": <cell_type>,
"parameters": {
<parameter_name>: <parameter_value>,
...
},
"attributes": {
<attribute_name>: <attribute_value>,
...
},
"port_directions": {
<port_name>: <"input" | "output" | "inout">,
...
},
"connections": {
<port_name>: <bit_vector>,
...
},
}
And <net_details> is:
{
"hide_name": <1 | 0>,
"bits": <bit_vector>
}
The "hide_name" fields are set to 1 when the name of this cell or net is
automatically created and is likely not of interest for a regular user.
The "port_directions" section is only included for cells for which the
interface is known.
Module and cell ports and nets can be single bit wide or vectors of multiple
bits. Each individual signal bit is assigned a unique integer. The <bit_vector>
values referenced above are vectors of this integers. Signal bits that are
connected to a constant driver are denoted as string "0" or "1" instead of
a number.
For example the following Verilog code:
module test(input x, y);
(* keep *) foo #(.P(42), .Q(1337))
foo_inst (.A({x, y}), .B({y, x}), .C({4'd10, {4{x}}}));
endmodule
Translates to the following JSON output:
{
"modules": {
"test": {
"ports": {
"x": {
"direction": "input",
"bits": [ 2 ]
},
"y": {
"direction": "input",
"bits": [ 3 ]
}
},
"cells": {
"foo_inst": {
"hide_name": 0,
"type": "foo",
"parameters": {
"Q": 1337,
"P": 42
},
"attributes": {
"keep": 1,
"src": "test.v:2"
},
"connections": {
"C": [ 2, 2, 2, 2, "0", "1", "0", "1" ],
"B": [ 2, 3 ],
"A": [ 3, 2 ]
}
}
},
"netnames": {
"y": {
"hide_name": 0,
"bits": [ 3 ],
"attributes": {
"src": "test.v:1"
}
},
"x": {
"hide_name": 0,
"bits": [ 2 ],
"attributes": {
"src": "test.v:1"
}
}
}
}
}
}
The models are given as And-Inverter-Graphs (AIGs) in the following form:
"models": {
<model_name>: [
/* 0 */ [ <node-spec> ],
/* 1 */ [ <node-spec> ],
/* 2 */ [ <node-spec> ],
...
],
...
},
The following node-types may be used:
[ "port", <portname>, <bitindex>, <out-list> ]
- the value of the specified input port bit
[ "nport", <portname>, <bitindex>, <out-list> ]
- the inverted value of the specified input port bit
[ "and", <node-index>, <node-index>, <out-list> ]
- the ANDed value of the specified nodes
[ "nand", <node-index>, <node-index>, <out-list> ]
- the inverted ANDed value of the specified nodes
[ "true", <out-list> ]
- the constant value 1
[ "false", <out-list> ]
- the constant value 0
All nodes appear in topological order. I.e. only nodes with smaller indices
are referenced by "and" and "nand" nodes.
The optional <out-list> at the end of a node specification is a list of
output portname and bitindex pairs, specifying the outputs driven by this node.
For example, the following is the model for a 3-input 3-output $reduce_and cell
inferred by the following code:
module test(input [2:0] in, output [2:0] out);
assign in = &out;
endmodule
"$reduce_and:3U:3": [
/* 0 */ [ "port", "A", 0 ],
/* 1 */ [ "port", "A", 1 ],
/* 2 */ [ "and", 0, 1 ],
/* 3 */ [ "port", "A", 2 ],
/* 4 */ [ "and", 2, 3, "Y", 0 ],
/* 5 */ [ "false", "Y", 1, "Y", 2 ]
]
Future version of Yosys might add support for additional fields in the JSON
format. A program processing this format must ignore all unknown fields.
\end{lstlisting}
\section{write\_smt2 -- write design to SMT-LIBv2 file}
\label{cmd:write_smt2}
\begin{lstlisting}[numbers=left,frame=single]
write_smt2 [options] [filename]
Write a SMT-LIBv2 [1] description of the current design. For a module with name
'<mod>' this will declare the sort '<mod>_s' (state of the module) and the
functions operating on that state.
The '<mod>_s' sort represents a module state. Additional '<mod>_n' functions
are provided that can be used to access the values of the signals in the module.
Only ports, and signals with the 'keep' attribute set are made available via
such functions. Without the -bv option, multi-bit wires are exported as
separate functions of type Bool for the individual bits. With the -bv option
multi-bit wires are exported as single functions of type BitVec.
The '<mod>_t' function evaluates to 'true' when the given pair of states
describes a valid state transition.
The '<mod>_a' function evaluates to 'true' when the given state satisfies
the asserts in the module.
The '<mod>_u' function evaluates to 'true' when the given state satisfies
the assumptions in the module.
The '<mod>_i' function evaluates to 'true' when the given state conforms
to the initial state.
-verbose
this will print the recursive walk used to export the modules.
-bv
enable support for BitVec (FixedSizeBitVectors theory). with this
option set multi-bit wires are represented using the BitVec sort and
support for coarse grain cells (incl. arithmetic) is enabled.
-mem
enable support for memories (via ArraysEx theory). this option
also implies -bv. only $mem cells without merged registers in
read ports are supported. call "memory" with -nordff to make sure
that no registers are merged into $mem read ports. '<mod>_m' functions
will be generated for accessing the arrays that are used to represent
memories.
-regs
also create '<mod>_n' functions for all registers.
-wires
also create '<mod>_n' functions for all public wires.
-tpl <template_file>
use the given template file. the line containing only the token '%%'
is replaced with the regular output of this command.
[1] For more information on SMT-LIBv2 visit http://smt-lib.org/ or read David
R. Cok's tutorial: http://www.grammatech.com/resources/smt/SMTLIBTutorial.pdf
---------------------------------------------------------------------------
Example:
Consider the following module (test.v). We want to prove that the output can
never transition from a non-zero value to a zero value.
module test(input clk, output reg [3:0] y);
always @(posedge clk)
y <= (y << 1) | ^y;
endmodule
For this proof we create the following template (test.tpl).
; we need QF_UFBV for this poof
(set-logic QF_UFBV)
; insert the auto-generated code here
%%
; declare two state variables s1 and s2
(declare-fun s1 () test_s)
(declare-fun s2 () test_s)
; state s2 is the successor of state s1
(assert (test_t s1 s2))
; we are looking for a model with y non-zero in s1
(assert (distinct (|test_n y| s1) #b0000))
; we are looking for a model with y zero in s2
(assert (= (|test_n y| s2) #b0000))
; is there such a model?
(check-sat)
The following yosys script will create a 'test.smt2' file for our proof:
read_verilog test.v
hierarchy -check; proc; opt; check -assert
write_smt2 -bv -tpl test.tpl test.smt2
Running 'cvc4 test.smt2' will print 'unsat' because y can never transition
from non-zero to zero in the test design.
\end{lstlisting}
\section{write\_smv -- write design to SMV file}
\label{cmd:write_smv}
\begin{lstlisting}[numbers=left,frame=single]
write_smv [options] [filename]
Write an SMV description of the current design.
-verbose
this will print the recursive walk used to export the modules.
-tpl <template_file>
use the given template file. the line containing only the token '%%'
is replaced with the regular output of this command.
THIS COMMAND IS UNDER CONSTRUCTION
\end{lstlisting}
\section{write\_spice -- write design to SPICE netlist file}
\label{cmd:write_spice}
\begin{lstlisting}[numbers=left,frame=single]
write_spice [options] [filename]
Write the current design to an SPICE netlist file.
-big_endian
generate multi-bit ports in MSB first order
(default is LSB first)
-neg net_name
set the net name for constant 0 (default: Vss)
-pos net_name
set the net name for constant 1 (default: Vdd)
-nc_prefix
prefix for not-connected nets (default: _NC)
-top top_module
set the specified module as design top module
\end{lstlisting}
\section{write\_verilog -- write design to Verilog file}
\label{cmd:write_verilog}
\begin{lstlisting}[numbers=left,frame=single]
write_verilog [options] [filename]
Write the current design to a Verilog file.
-norename
without this option all internal object names (the ones with a dollar
instead of a backslash prefix) are changed to short names in the
format '_<number>_'.
-noattr
with this option no attributes are included in the output
-attr2comment
with this option attributes are included as comments in the output
-noexpr
without this option all internal cells are converted to Verilog
expressions.
-blackboxes
usually modules with the 'blackbox' attribute are ignored. with
this option set only the modules with the 'blackbox' attribute
are written to the output file.
-selected
only write selected modules. modules must be selected entirely or
not at all.
\end{lstlisting}
Reference in New Issue View Git Blame Copy Permalink