Recovering spectre simulations

H

Hristo Brachkov

Guest
Hello, I have the following question:


Sometimes we are running long PSS, PNOISE and TRAN simulations. Of course
sometimes Cadence or the connection to the server crashes and all
simulations results are lost.

Could anybody tell me if there is a way to set the things so that after an
interruption of the simulations one would be able to restart the simulations
from the "crash point" ?

Thanks,
Hristo
 
If you have an idea of where its crashing, you could save the states
at periodic intervals and restart from the save state. I use this
save-restart SST feature of High Q oscillator, which might otherwise
have convergence problems.
Partha


"Hristo Brachkov" <hristob@csPOIS.tut.fi> wrote in message news:<br9u6v$2mr$1@news.cc.tut.fi>...
Hello, I have the following question:


Sometimes we are running long PSS, PNOISE and TRAN simulations. Of course
sometimes Cadence or the connection to the server crashes and all
simulations results are lost.

Could anybody tell me if there is a way to set the things so that after an
interruption of the simulations one would be able to restart the simulations
from the "crash point" ?

Thanks,
Hristo
Click to expand...
 
Partha, could you tell me how exactly you are doing it?


"Partha" <cadeguy@yahoo.com> wrote in message
news:7f58031f.0312131245.15aebf86@posting.google.com...
If you have an idea of where its crashing, you could save the states
at periodic intervals and restart from the save state. I use this
save-restart SST feature of High Q oscillator, which might otherwise
have convergence problems.
Partha


"Hristo Brachkov" <hristob@csPOIS.tut.fi> wrote in message
news:<br9u6v$2mr$1@news.cc.tut.fi>...
Hello, I have the following question:


Sometimes we are running long PSS, PNOISE and TRAN simulations. Of
course
sometimes Cadence or the connection to the server crashes and all
simulations results are lost.

Could anybody tell me if there is a way to set the things so that after
an
interruption of the simulations one would be able to restart the
simulations
from the "crash point" ?

Thanks,
Hristo
Click to expand...
 
Hristo,
I have used this feature in Oscillator(driven) circuits. readpss and
writepss are options for the steady state analysis to save the results
of PSS so one may not have to rerun tha analysis or for further
PAC/Pnoise analysis.
High Q factor oscillator sometimes do not converge, so i resort to
incrementally saving the PSS states increasing the Q and rerunning it
again with the saved state.

But, I am pretty confident this cannot be used for recovering from
crashed simulations.:-(

If you are using versions 4.4.5 or lower, i know that spectre had an
"issue" with the 2Gb swap file limit. But in future versions this was
fixed by generating a series of files as soon as the limit was
reached.

Partha


"Hristo Brachkov" <hristob@SPAM.REMOVEcs.tut.fi> wrote in message news:<brik2o$rsc$1@news.cc.tut.fi>...
Partha, could you tell me how exactly you are doing it?


"Partha" <cadeguy@yahoo.com> wrote in message
news:7f58031f.0312131245.15aebf86@posting.google.com...
If you have an idea of where its crashing, you could save the states
at periodic intervals and restart from the save state. I use this
save-restart SST feature of High Q oscillator, which might otherwise
have convergence problems.
Partha


"Hristo Brachkov" <hristob@csPOIS.tut.fi> wrote in message
news:<br9u6v$2mr$1@news.cc.tut.fi>...
Hello, I have the following question:


Sometimes we are running long PSS, PNOISE and TRAN simulations. Of
course
sometimes Cadence or the connection to the server crashes and all
simulations results are lost.

Could anybody tell me if there is a way to set the things so that after
an
interruption of the simulations one would be able to restart the
simulations
from the "crash point" ?

Thanks,
Hristo
Click to expand...
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top