Product Documentation
Spectre Circuit Simulator Reference
Product Version 23.1, September 2023

Deferred Set Options (set)

Description

The deferred set options statement sets or changes various program control options. You can set the options in any order and, once set, the options retain their value until reset. The set statement is queued with all analyses and is executed sequentially (The changes made to these options are deferred until the statement setting them is encountered). To set temp, tnom, scalem, or scale, use the alter statement. For further options, see individual analyses.

Syntax

Name set parameter=value ...

Parameters

Tolerance parameters

reltol

0.001

Relative convergence criterion.

residualtol

1.0

Tolerance ratio for residual (multiplies reltol).

vabstol

1.0e-6 V

Voltage absolute tolerance convergence criterion.

iabstol

1.0e-12 A

Current absolute tolerance convergence criterion.

Temperature parameters

tempeffects

all

Temperature effect selector. If tempeffect = vt, only thermal voltage varies with temperature; if tempeffect = tc, parameters that start with tc are active and thermal voltage is dependent on temperature; and if tempeffect = all, all built-in temperature models are enabled. Possible values are vt, tc and all.

Convergence parameters

homotopy

all

Method used when no convergence on initial attempt of DC analysis. Users can specify methods and their orders by giving vector setting such as homotopy=[source ptran gmin]. Possible values are none, gmin, source, dptran, ptran, all, arclength and tranrampup.

newton

normal

Method used on DC analysis. Users can specify methods newton=[none | normal], and the default value is normal.. Possible values are none and normal.

limit

dev

Limiting algorithms to aid DC convergence. Possible values are delta, log, dev and l2.

gmethod

dev

Stamp gdev, gnode or both in the homotopy methods (other than dptran). See below for more information. Possible values are dev, node and both.

dptran_gmethod

node

Stamp gdev, gnode or both in the dptran (homotopy) methods. Possible values are dev, node and both.

try_fast_op

yes

This feature often speeds up the DC solution. For hard to converge designs, this feature fails and other methods are applied. In corner cases, this feature may have negative effects. If the DC analysis is unusually slow or if the memory usage of various processes keeps growing or if DC gets stuck even before homotopy methods start, try setting this option to no.. Possible values are no and yes.

Component parameters

compatible

spectre

Encourage device equations to be compatible with a foreign simulator. This option does not affect input syntax. Possible values are spectre, spice2, spice3, cdsspice, hspice, spiceplus, eldo, sspice, mica, tispice and pspice.

approx

no

Use approximate models. Difference between approximate and exact models is generally very small. Possible values are no and yes.

Error-checking parameters

diagnose

no

Print additional information that might help diagnose accuracy and convergence problems. Possible values are no and yes.

opptcheck

yes

Check operating point parameters against soft limits. Possible values are no and yes.

Resistance parameters

gmin

1e-12 S

Minimum conductance across each nonlinear device.

gmin_check

max_v_only

Specifies that effect of gmin should be reported if significant. Possible values are no, max_v_only, max_only and all.

gmindc

1e-12 S

Minimum conductance across each nonlinear device in DC analysis. If gmindc is not given explicitly, the value of gmindc will be equal to gmin.Default value is 1.0e-12.

rforce

1

Resistance used when forcing nodesets and node-based initial conditions.

Quantity parameters

quantities

no

Print quantities. If quantities=min, the simulator will print out all defined quantities; if quantities=full, the simulator will also print a list of nodes and their quantities. Possible values are no, min and full.

Annotation parameters

narrate

yes

Narrate the simulation. Possible values are no and yes.

annotate

no

Degree of annotation. Possible values are no and title.

debug

no

Give debugging messages. Possible values are no and yes.

info

yes

Give informational messages. Possible values are no and yes.

note

yes

Give notice messages. Possible values are no and yes.

maxnotes

5

Maximum number of times a notice is issued per analysis. Note that this option has no effect on notices issued as part of parsing the netlist. Use the -maxnotes command-line option to control the number of all notices issued.

warn

yes

Give warning messages. Possible values are no and yes.

maxwarns

5

Maximum number of times a warning message is issued per analysis. Note that this option has no effect on warnings issued as part of parsing the netlist. Use the -maxwarns command-line option to control the number of all warnings issued.

maxwarnstologfile

5

Maximum number of times a warning message is printed to the log file per analysis. Note that this option has no effect on warnings printed as part of parsing the netlist. Use the -maxwarnstolog command-line option to control the number of all warnings printed to the log file.

maxnotestologfile

5

Maximum number of times a notice message is printed to the log file per analysis. Note that this option has no effect on notices printed as part of parsing the netlist. Use the -maxnotestolog command-line option to control the number of all notices printed to the log file.

error

yes

Generates error messages. Possible values are no and yes.

digits

5

Number of digits used when printing numbers.

measdgt

0

Number of decimal digits in floating point numbers in measurement output in mt0 format.

notation

eng

The notation to be used for displaying real numbers to the screen. Possible values are eng, sci and float.

Matrix parameters

pivotdc

no

Use numeric pivoting on every iteration of DC analysis. Possible values are no and yes.

pivrel

0.001

Relative pivot threshold.

pivabs

0

Absolute pivot threshold.

preorder

partial

Try this option when simulation runs out of memory or if the simulation is unreasonably slow for the size of your design. It controls the amount of matrix preordering that is done and may lead to much fewer matrix fill-ins in some cases. Known cases are: designs with very large number of small resistors or large number of behavioral instances containing voltage based equations.. Possible values are partial, full and limited.

limit_diag_pivot

yes

If set to yes, there is a limit on the number of matrix fill-ins when selecting a pivot from a diagonal. For backward compatibility set this to no.. Possible values are no and yes.

rebuild_matrix

no

If yes, rebuild circuit matrix at the beginning of ac, acmatch, dc, dcmatch, montecarlo, pz, stb, sweep, tdr, and tran analyses. This is to ensure consistent matrix ordering at the beginning of the analyses for consistent results. Notice that rebuild circuit matrix can incur performance overhead. Possible values are no and yes.

icversion

1

Convert initial condition to initial guess, when .ic statements exist in netlist and there are no other options to set IC or nodeset.

minstepversion

1

Minstep algorithm flow control. If minstepversion=0, desperate big jump is taken when minstep is reached. If minstepversion=1, a forward/backward search algorithm is taken to find a converged solution at small step size. By default, minstepversion=1.

try_hard_in_disaster

yes

When minstepversion=1, this option means whether simulator should try hard to search for a converged solution in disaster stage. Possible values are no and yes.


Return to top
 ⠀
X