spectreS: arithmetic exception

spectreS: arithmetic exception

Post by Luigi Caponett » Fri, 12 Feb 1999 04:00:00



Could anyone please help me with this error found by spectreS in IC
analysis?

*********** spectre.out ***********

spectre (ver. 4.4.1.78 -- 20 Jan 1997).

Simulating `raw/TESTshifter256' on sally.ct.infn.it at 4:49:30 PM, Thur
Feb 11,
        1999.

Circuit inventory:
              nodes 1062
          equations 4269
              bsim3 1601  
          capacitor 6567  
           resistor 1    
            vsource 5    

******************************************************
Transient Analysis `timeSweep': time = (0 s -> 200 ns)
******************************************************

Fatal error found by spectre during IC analysis, during transient
analysis
        `timeSweep'.
    Arithmetic exception.

Aggregate audit (4:49:42 PM, Thur Feb 11, 1999):
Time used: CPU = 11.5 s, elapsed = 12 s, util. = 95.5%.
Virtual memory used = 9.45 Mbytes.
spectre completes with 1 error, 0 warnings, and 0 notices.
spectre terminated prematurely due to fatal error.

*********** end spectre.out ***********

This is the output from a 100n transient simulation.
The problem seems to be related to the IC calculation.
I tried tuning the convergence parameters and even to made a DC
simulation, with no results.

Every help will be appreciated cause we have an urgent need to complete
this simulation.

TIA

luigi

--


Universita' di Catania           | +39-095-7195-264 [v] |  
95129 Catania - ITALY            | +39-095-383023   [f] |

 
 
 

spectreS: arithmetic exception

Post by Karim EMBARE » Sun, 14 Feb 1999 04:00:00


Hi !

        Already got this message in 4.3 version of spectreS in 1995.
I have reported to Cadence, but I never had serious feedback. They said
to me that it could come from operating system or lack of memory, or
some other things like this.

        Sure, it was not completely false because the problem was
random and sometimes,  re-booting workstation (sparc station 5 under
sunOs at this time) gave results.

        Sorry, to cannot help you more !



> Could anyone please help me with this error found by spectreS in IC
> analysis?

> *********** spectre.out ***********

> spectre (ver. 4.4.1.78 -- 20 Jan 1997).

> Simulating `raw/TESTshifter256' on sally.ct.infn.it at 4:49:30 PM, Thur
> Feb 11,
>         1999.

> Circuit inventory:
>               nodes 1062
>           equations 4269
>               bsim3 1601
>           capacitor 6567
>            resistor 1
>             vsource 5

> ******************************************************
> Transient Analysis `timeSweep': time = (0 s -> 200 ns)
> ******************************************************

> Fatal error found by spectre during IC analysis, during transient
> analysis
>         `timeSweep'.
>     Arithmetic exception.

> Aggregate audit (4:49:42 PM, Thur Feb 11, 1999):
> Time used: CPU = 11.5 s, elapsed = 12 s, util. = 95.5%.
> Virtual memory used = 9.45 Mbytes.
> spectre completes with 1 error, 0 warnings, and 0 notices.
> spectre terminated prematurely due to fatal error.

> *********** end spectre.out ***********

> This is the output from a 100n transient simulation.
> The problem seems to be related to the IC calculation.
> I tried tuning the convergence parameters and even to made a DC
> simulation, with no results.

> Every help will be appreciated cause we have an urgent need to complete
> this simulation.

> TIA

> luigi

> --


> Universita' di Catania           | +39-095-7195-264 [v] |
> 95129 Catania - ITALY            | +39-095-383023   [f] |


 
 
 

spectreS: arithmetic exception

Post by Andrew Becke » Mon, 15 Feb 1999 04:00:00


Unfortunately, it's not really possible to be precise without the
circuit, but I've seen a similar problem reported with bsim3 recently.
In fact it may even have been the same problem, reported via your
support channels?

The problem may be due to something not being right in the circuit,
but this may be causing problems in the sparse matrix package.

Try setting "homotopy=dptran" in the DC options, and also set
pivrel=0.5 and pivotdc=yes in the simulator options.

This should make it converge faster too (if it's the problem I'm thinking
of). If this doesn't help, report this to Cadence through your normal support
channels.

Best Regards,

Andrew.



>Could anyone please help me with this error found by spectreS in IC
>analysis?

>*********** spectre.out ***********

>spectre (ver. 4.4.1.78 -- 20 Jan 1997).

>Simulating `raw/TESTshifter256' on sally.ct.infn.it at 4:49:30 PM, Thur
>Feb 11,
>        1999.

>Circuit inventory:
>              nodes 1062
>          equations 4269
>              bsim3 1601  
>          capacitor 6567  
>           resistor 1    
>            vsource 5    

>******************************************************
>Transient Analysis `timeSweep': time = (0 s -> 200 ns)
>******************************************************

>Fatal error found by spectre during IC analysis, during transient
>analysis
>        `timeSweep'.
>    Arithmetic exception.

>Aggregate audit (4:49:42 PM, Thur Feb 11, 1999):
>Time used: CPU = 11.5 s, elapsed = 12 s, util. = 95.5%.
>Virtual memory used = 9.45 Mbytes.
>spectre completes with 1 error, 0 warnings, and 0 notices.
>spectre terminated prematurely due to fatal error.

>*********** end spectre.out ***********

>This is the output from a 100n transient simulation.
>The problem seems to be related to the IC calculation.
>I tried tuning the convergence parameters and even to made a DC
>simulation, with no results.

>Every help will be appreciated cause we have an urgent need to complete
>this simulation.

>TIA

>luigi

>--


>Universita' di Catania           | +39-095-7195-264 [v] |  
>95129 Catania - ITALY            | +39-095-383023   [f] |

--
*************************************************************
* Andrew Beckett               * Tel:   +44 1344 360333     *
* Technical Leader, Custom IC  * Fax:   +44 1344 360324     *

* Bagshot Road                 *                            *
* Bracknell.  RG12 3PH         *                            *
*************************************************************