ocean script - input.scs copy

In IC613 you have OCEAN XL which can do this - it uses ADE XL's history mechanism, so you can see the setup from previous runs, including the netlists.

I think in standard OCEAN scripts (in IC5141 or IC613) you can probably do this:

envSetVal("asimenv.misc" "numberOfSavedRuns" 'int 5)

which will auto-save the simulation results (and netlists) for the last five runs. It does this in ADE, and I believe it would also affect OCEAN (although I have not tried it - bit busy today).

Regards,

Andrew. 

 Hi

 Unfortunately mine is still IC5 series. I tried  envSetVal("asimenv.misc" "numberOfSavedRuns" 'int 5) option in my script and it surely copied 5 previous psf folder but not netlist nor input.scs.

Is there any other way to copy netlist every single simulation?

 Thanks,

 Hi aplumb,

 Thanks for your info. 

Quite unfortunate but error message came up as followed when I tried via GUI ADE.

------------------------------

LBS_CLUSTER_MASTER is not set. Checking DP for LSF setup....

ERROR (209): The DP environment variables are not set.
To use the LBS mode, set LBS_CLUSTER_MASTER to the name of the cluster
master host. To use the LSF mode, set LBS_BASE_SYSTEM to LBS_LSF 

------------------------------

It sounds I need to talk with admin guy but is there any other way to avoid such a painful time?

Thanks,

Really? I just tried this in IC5141:

envSetVal("asimenv.misc" "numberOfSavedRuns" 'int 5)
simulator( 'spectre )
design(  "./simulation/ampTest/spectre/schematic/netlist/netlist")
path( "./Models" )
modelFile(
    '("myModels.scs" "")
)
analysis('ac ?start "100"  ?stop "150M"  ?dec "20"  )
analysis('tran ?stop "3u"  )
analysis('sens ?analyses_list list("dcOp")  ?net list("/out")  )
analysis('dc ?saveOppoint t  )
desVar(   "CAP" .8p     )
temp( 27 )
run()
selectResult( 'tran )
plot(getData("/vin") getData("/out") )
selectResult( 'ac )
plot(getData("/vin") getData("/out") )

And after the simulation, I got a "1" directory, which as you can see has the input.scs in it

andrewb_31> ls -lG {1,schematic}/netlist/input.scs
-rw-r--r-- 1 andrewb 2280 Sep 23 11:08 1/netlist/input.scs
-rw-r--r-- 1 andrewb 2280 Sep 23 11:08 schematic/netlist/input.scs

So it appears to work for me!

As for the distributed processing option, well, you need to set up distributed processing. It won't work without any setup - I suggest you read the manual on this! <ICinstDir>/doc/distproc/distproc.pdf

Regards,

Andrew.

Hi Andrew, 

Yes there is numbered folder as you mentioned, but no "netlist" folder in it apparantly as you can see here.

tklin13..> pwd
/MydfII/simulation/sim_osc2_gmbias/spectre/schematic/org/32
tklin13..> lt
total 7
drwxrwsr-x    3 aaaaaaaa eng  512 Sep 23 23:55 ./
drwxrwsr-x    2 aaaaaaaa eng  512 Sep 24 00:06 psf/
drwxrwsr-x  137 aaaaaaaa eng 5120 Sep 24 00:06 ../

I think I am close to complete but something ( or some environmental setting) is missing.  

Thanks very much.

I wonder if this is because your directory hierarchy has the extra "org" directory in it? Normally the directory structure would be:

/MydfII/simulation/sim_osc2_gmbias/spectre/schematic - containing netlist and psf dirs

and then the new numbered dir would be:

/MydfII/simulation/sim_osc2_gmbias/spectre/32 - containing netlist and psf dirs

i.e. it copies the "view" directory.

It may be that the non-ADE-like directory structure is confusing things?

Regards,

Andrew.

 Hi Andrew,

 It seems not.

I also tried without "org" but no luck. Similarly only psf folder was generated. I have no clue with this.

Thanks,

 I'm not explicitly answering your question, but if you were willing to run your simulation entirely within Ocean, you can send the simulation results wherever you like with:

  design("LibName" "CellName" "viewname" "r")
  netlist_dir =strcat(envGetVal("asimenv.startup" "projectDir") "/" "CellName" "/spectre/" "ViewName" "/netlist")
  res_dir = strcat(envGetVal("asimenv.startup" "projectDir") "/" "CellName" "/spectre/" "ViewName")
  netlistDir(netlist_dir) 
  resultsDir(res_dir) ; set the results directory for the run

With this method, you can have a different results directory for each simulation type.  You're still stuck with having to explicitly save/move your results if you want to save them from run to run, but you don't have all of your simulations going into one directory.