Save transient values from a Monte Carlo simulation to a .txt file

Sorry, I made a mistake and cant edit the post, the file names are different,

Out_It1 = outfile(".../Iteration_1.txt" "w")
Out_It2 = outfile(".../Iteration_2.txt" "w")

.... 

Rest is the same.

Dear Yasso,

Yasso said:

Currently I have an schematic which I need to simulate and save/retrieve specific transient values. My problem is that it doesnt save all the iterations,

In the Monte Carlo GUI, did you happen to check the Save Waveforms radio button as shown in Figure 1 below?

Yasso said:

for(i 1 19200
sprintf(g "/I%d/Vp" i)
ocnPrint(?output Out_It1  value(VT(g) 100m 'iteration 1) ?numberNotation 'none)
)

Are you trying to access a signal defined as "/IXX/Vp" where XX=[1,19200]? I'm a bit confused as I was assuming the 19200 was the number of Monte Carlo iterations.

Yasso said:

Is it possible to do using an ocean script

Definitely. I have accessed any Monte Carlo data I want from a set of Monte Carlo simulations using ocean. My methodology differs from yours as I traverse the Monte Carlo data base using ocean to access the data for each iteration for further post-processing.

Shawn

Figure 1

Yes, I have the "Save Waveforms" enabled to true,

ocnxlMonteCarloOptions( ?mcNumPoints "2" ?mcYieldTarget "99.865" ?mcNumBins "" ?samplingMode "random" ?wcdMinYieldSigma "3.0" ?dumpParamMode "yes" ?mcYieldAlphaLimit "0.05" ?mcStartingRunNumber "1" ?mcSigmaTarget "3" ?monteCarloSeed "12345" ?sssNumPoints "7000" ?highYieldExpectedYieldSigma "3" ?mcMethod "all" ?wcdInitialPoints "200" ?wcdMaxIterations "10" ?saveAllPlots "1" ?highYieldUseHistoryMC "0" ?saveProcess "1" ?wcdAutoNumPoints "1" ?mcStopEarly "0" ?highYieldCreateCorners "0" ?saveMismatch "1" ?wcdSkipLowYieldSpecs "1" ?wcdEnableVarReduction "1" )

The code lines I added are writting into the .txt file the voltage values from instances with the name "I<number of the instance>", exactly the parrameter "Vp". It's basically an array of instances. So for each iteration in MC there are 19000 Vp values that need to be saved into a .txt. Unfortunately it only saves the first or the last iteration, not both in separated files. Just to clarify, I would like to simulate 30 iterations of this array composed of 19000 elements and save the Vp value in different files. This will later be used in a Matlab/Phyton script.

I'm using a version of cadence 6.1.8-64b (2018-19)

And thanks for the help!

Yass

Dear Yass,

Thank for clarifying your simulation! I now understand that you do indeed want to save each of the 19200 individual node voltages for each monte carlo iteration in a separate file.

Yasso said:

Just to clarify, I would like to simulate 30 iterations

Perhaps I am overlooking something obvious, but it appears you are only specifying 2 iterations - not 30 - in your oceanXL command:

Yasso said:

ocnxlMonteCarloOptions( ?mcNumPoints "2" ?mcYieldTarget "99.865" ?mcNumBins "" ?samplingMode "random" ?wcdMinYieldSigma "3.0" ?dumpParamMode "yes"

Note that ?mcNumPoints is set to "2".

1. Hence, it seems there will only be two iterations and not 30 iterations. As such, you will only be able to generate two files with 19200 values in each - one for the first iteration and a second for the second iteration.

Yasso said:

Unfortunately it only saves the first or the last iteration, not both in separated files. Just to clarify, I would like to simulate 30 iterations of this array composed of 19000 elements and save the Vp value in different files.

2. Your ocnPrint() command syntax may be the reason you are finding your result is only for the first or last iteration. The syntax you are using does not appear to include sufficient information to access the iteration you are interested in saving. oceanXL is the language to use with Assembler and Explorer to take advantage of their database structures. I can provide an example ocean script (not oceanXL) that will traverse your Monte Carlo database and access each voltage and place it in a text file. However, there is a recommended method using oceanXL in the Troubleshooting Article at URL:

https://support.cadence.com/apex/ArticleAttachmentPortal?id=a1Od0000002JnSHEA0&pageName=ArticleContent

The article provides a SKILL routine CCSDumpFile() to include in your .csdinit file in your working directory.  You define your output expressions using the CCDDumpFile() function.

If you would rather run a conventional ocean script and are interested in an example, I placed an example at the Forum post at URL:

https://community.cadence.com/cadence_technology_forums/f/custom-ic-skill/54157/ocean-script-to-get-monte-carlo-iteration-id-directory

Shawn

Hello Shawn, sorry for the late reply,

Yes, the num of iterations is set to 2 for testing. and I will change it to higher ammounts the moment it works. ocnPrint() was my default command for printing data to .txt, but now knowing that the databases are constructed different for Monte Carlo It explains why I dont get the results I wanted.

I can't see the support.cadence links because where I work I need to ask EuroPractice (the middle man between Cadence and our work enviroment) to send the tickets for me. I'm checking the seccond script example you have linked me, but it's quite advanced.

Is it possible for an example with a structure like this,

;====================Set to Maestro mode explorer =============================
ocnSetXLMode("explorer")
ocnxlProjectDir( "/simulation" )
ocnxlTargetCellView( ""ARRAY" "ARRAY_TEST" "MC" )
ocnxlResultsLocation( "" )
ocnxlSimResultsLocation( "" )
ocnxlMaxJobFail( 20 )

;====================== Tests setup ============================================


ocnxlBeginTest("ARRAY_TEST:1")
simulator( 'spectre )
design( "ARRAY" "ARRAY_TEST" "schematic")
modelFile( 
    '("../corner.lib.scs" "mc")

)
analysis('tran ?stop "101m" ?errpreset "conservative"  )

desVar(	  "sigma" 3	)
envOption(
	'analysisOrder  list("pz" "dcmatch" "stb" "tran" "envlp" "ac" "dc" "lf" "noise" "xf" "sp" "pss" "pac" "pstb" "pnoise" "pxf" "psp" "qpss" "qpac" "qpnoise" "qpxf" "qpsp" "hb" "hbac" "hbstb" "hbnoise" "hbxf" "sens" "acmatch") 
)
option(	'pzSeverity  "None" 
	'noiseSeverity  "None" 
	'spSeverity  "None" 
	'acSeverity  "None" 
	'dcOpSeverity  "None" 
	'dcSeverity  "None" 
	'tranSeverity  "None" 
	'gmin  "0" 
)
option(	'dc_pivot_check  "yes"  
	'gmin  "1e-18" 
)
option( ?categ 'turboOpts
	'apsplus  t 
	'errorLevel  "Moderate" 
	'uniMode  "APS" 
)
temp( 27 ) 
Out_It1 = outfile("/It1.txt" "w")
Out_It2 = outfile("/It2.txt" "w")
ocnxlEndTest() 

;====================== Sweeps setup ===========================================

;====================== Model Group setup ==========================================

;====================== Corners setup ==========================================

;====================== Checks and Asserts setup ============================================
ocnxlPutChecksAsserts(?netlist nil)

;====================== Job setup ==============================================
ocnxlJobSetup( '(
	"blockemail" "1"
	"configuretimeout" "300"
	"defaultcpuvalue" "1"
	"defaultmemoryvalue" "1000"
	"distributionmethod" "Local"
	"jobruntype" "ICRP"
	"estimatememoryvalue" ""
	"estimationsimulationmode" "0"
	"lingertimeout" "300"
	"maxjobs" "1"
	"name" "Maestro Default"
	"preemptivestart" "1"
	"providecpuandmemorydata" "1"
	"reconfigureimmediately" "1"
	"runpointsvalue" "5"
	"runtimeout" "-1"
	"scaleestimatedbycpu" "100"
	"scaleestimatedbymemory" "100"
	"showerrorwhenretrying" "1"
	"showoutputlogerror" "0"
	"startmaxjobsimmed" "1"
	"starttimeout" "300"
	"usesameprocess" "1"
	"warndisklow" "0"
	"warnthresholdvalue" "100"
) )

;====================== Disabled items =========================================
ocnxlSetAllParametersDisabled(t)

;====================== Run Mode Options ======================================
ocnxlMonteCarloOptions( ?mcNumPoints "2" ?mcYieldTarget "99.865" ?mcNumBins "" ?samplingMode "random" ?wcdMinYieldSigma "3.0" ?dumpParamMode "yes" ?mcYieldAlphaLimit "0.05" ?mcStartingRunNumber "1" ?mcSigmaTarget "3" ?monteCarloSeed "12345" ?sssNumPoints "7000" ?highYieldExpectedYieldSigma "3" ?mcMethod "all" ?wcdInitialPoints "200" ?wcdMaxIterations "10" ?saveAllPlots "1" ?highYieldUseHistoryMC "0" ?saveProcess "1" ?wcdAutoNumPoints "1" ?mcStopEarly "0" ?highYieldCreateCorners "0" ?saveMismatch "1" ?wcdSkipLowYieldSpecs "1" ?wcdEnableVarReduction "1" )

;====================== Starting Point Info ======================================

;====================== Run command ============================================
ocnxlRun( ?mode 'monteCarlo ?nominalCornerEnabled t ?allCornersEnabled nil ?allSweepsEnabled t)
ocnxlOutputSummary(?yieldSummary t ?exprSummary nil ?specSummary nil ?detailed nil)
ocnxlOpenResults()

< CODE TO EXTRACT DATA FROM THE MC ITERATIONS >
-----
-----
-----
-----
;====================== End XL Mode command ===================================
ocnxlEndXLMode("explorer")
close(Out_Tig_It1)
close(Out_Tig_It2)

Where in the section below a code/line of code to extract the data?

Sorry if I'm asking something complicated, I know it requires work, If theres an ocnXL document with the informations on how to extract this I will gladly take it.

Thanks again, and happy holidays!

Yass 

Dear Yass,

Yasso said:

Yes, the num of iterations is set to 2 for testing. and I will change it to higher ammounts the moment it works.

I now understand and thank you, Yass, for the clarification!

Yasso said:

I can't see the support.cadence links because where I work I need to ask EuroPractice (the middle man between Cadence and our work enviroment) to send the tickets for me.

That is unfortunate - please continue to try to access the documents. I understand that I am not permitted to include the content of an On-line support document in a Forum post and hence, with apologies, I cannot provide you the document I referred to.

Yasso said:

I'm checking the seccond script example you have linked me, but it's quite advanced.

It really is not - otherwise I would not be able to understand it!! I tried to remove all the unnecessary code that illustrate a specific example and include the code below as a text file. The lines you need to modify are highlighted in yellow in Figure 1. You may still choose not to try this, but I wanted to at least try to make it appear less complicated to you.

Yasso said:

Where in the section below a code/line of code to extract the data?

Unless I am misunderstanding your question, the code to extract the data is where the text "< CODE TO EXTRACT DATA FROM THE MC ITERATIONS >" appears. This is after the run command and is preceded by the command to open the results database.

Shawn

; Interative directory number (set to value X of your Montecarlo.X results directory, in the case X=0)

montecarlo_num = 0
sprintf(montecarlo_num_val "%d" montecarlo_num)

; Main routine

simulator('spectre)
date = strcat("040822")
circuit_name = strcat("<name_of_your_test>")
base_results=strcat("<full_UNIX_path_to_your_simulation_maestro_directory>/results/maestro/MonteCarlo." montecarlo_num_val "/")
sub_dir = strcat("/<name_of_your_test>/groupRunDataDir/psf/mc1_separate/")
sub_netlist_dir = strcat("/<name_of_your_test>/groupRunDataDir/")

sprintf(montecarlo_num_val "%d" montecarlo_num)

init_val = 1
final_val = 200 ;set final_value = 2 in your example case or 30 in your final case

sim_case = init_val

; When you run a Monte Carlo simulation, the netlists are shared between simulations
; the ocean script runs two loops to traverse this data base. sim_case_delta is the number
; of simulations that share a common netlist. The value can be found by examining the number of directories
; under the /<name_of_your_test>/groupRunDataDir/psf/mc1_separate/ directory in your results path

sim_case_delta = 3 

while( sim_case < (final_val + 1)

printf("sim_case = %d\n" sim_case)
sprintf(sim_case_val "%d" sim_case)

sub_sim_case = sim_case

while( (sub_sim_case < (sim_case + sim_case_delta)) && (sub_sim_case < (final_val + 1))

;; -----------
sprintf( sub_sim_case_val "%03d" sub_sim_case)
printf( "sub_sim_case = %03d\n" sub_sim_case)

results=strcat( base_results sim_case_val sub_dir sub_sim_case_val )
printf("Results directory\n%s\n" results)

netlist_dir = strcat( base_results sim_case_val sub_netlist_dir "netlist")
printf("Netlist directory\n%s\n" netlist_dir)
;; -----------

openResults(results)
selectResult('tran)
;ocnPrint(?output "outputs.txt" outputs())


printf("Completed reading waveforms.\n")


if((sub_sim_case == init_val) then

; Define output file for simulation results and write header

sprintf(fname "<name_of_your_output_file")
fp_params = outfile( fname "w")
fprintf(fp_params "Corner number,Process,vdda (mV),Temperature (C),") ; header for output file
; Place  output expressions here to add to file
close(fp_params)
else

fp_params = outfile( fname "a")

; Place same output expressions here to add to file
close(fp_params)

)

sub_sim_case = sub_sim_case + 1) ; close sub_sim_case loop

sim_case = sim_case + sim_case_delta) ; close sim_case loop

exit()

simplified_annotated_mc_params_sml_excel_122622.xlsx

Dear Yass,

I tried to respond to each of your concerns, but my response was flagged as spam. Hopefully, it will be re-classified for you to view it. Sorry, but I'm  to sure I understand what triggered its "spam rating".

The text of my response follows.

Dear Yass,

Yasso said:
Yes, the num of iterations is set to 2 for testing. and I will change it to higher ammounts the moment it works.
I now understand and thank you, Yass, for the clarification!

Yasso said:
I can't see the support.cadence links because where I work I need to ask EuroPractice (the middle man between Cadence and our work enviroment) to send the tickets for me.
That is unfortunate - please continue to try to access the documents. I understand that I am not permitted to include the content of an On-line support document in a Forum post and hence, with apologies, I cannot provide you the document I referred to.

Yasso said:
I'm checking the seccond script example you have linked me, but it's quite advanced.
It really is not - otherwise I would not be able to understand it!! I tried to remove all the unnecessary code that illustrate a specific example and include the code below as a text file. The lines you need to modify are highlighted in yellow in Figure 1. You may still choose not to try this, but I wanted to at least try to make it appear less complicated to you.

Yasso said:
Where in the section below a code/line of code to extract the data?
Unless I am misunderstanding your question, the code to extract the data is where the text "< CODE TO EXTRACT DATA FROM THE MC ITERATIONS >" appears. This is after the run command and is preceded by the command to open the results database.

Shawn

Figure 1

Document: simplified_annotated_mc_params_sml_122622.txt

; Interative directory number (set to value X of your Montecarlo.X results directory, in the case X=0)

montecarlo_num = 0
sprintf(montecarlo_num_val "%d" montecarlo_num)

; Main routine

simulator('spectre)
date = strcat("040822")
circuit_name = strcat("<name_of_your_test>")
base_results=strcat("<full_UNIX_path_to_your_simulation_maestro_directory>/results/maestro/MonteCarlo." montecarlo_num_val "/")
sub_dir = strcat("/<name_of_your_test>/groupRunDataDir/psf/mc1_separate/")
sub_netlist_dir = strcat("/<name_of_your_test>/groupRunDataDir/")

sprintf(montecarlo_num_val "%d" montecarlo_num)

init_val = 1
final_val = 200 ;set final_value = 2 in your example case or 30 in your final case

sim_case = init_val

; When you run a Monte Carlo simulation, the netlists are shared between simulations
; the ocean script runs two loops to traverse this data base. sim_case_delta is the number
; of simulations that share a common netlist. The value can be found by examining the number of directories
; under the /<name_of_your_test>/groupRunDataDir/psf/mc1_separate/ directory in your results path

sim_case_delta = 3 

while( sim_case < (final_val + 1)

printf("sim_case = %d\n" sim_case)
sprintf(sim_case_val "%d" sim_case)

sub_sim_case = sim_case

while( (sub_sim_case < (sim_case + sim_case_delta)) && (sub_sim_case < (final_val + 1))

;; -----------
sprintf( sub_sim_case_val "%03d" sub_sim_case)
printf( "sub_sim_case = %03d\n" sub_sim_case)

results=strcat( base_results sim_case_val sub_dir sub_sim_case_val )
printf("Results directory\n%s\n" results)

netlist_dir = strcat( base_results sim_case_val sub_netlist_dir "netlist")
printf("Netlist directory\n%s\n" netlist_dir)
;; -----------

openResults(results)
selectResult('tran)
;ocnPrint(?output "outputs.txt" outputs())


printf("Completed reading waveforms.\n")


if((sub_sim_case == init_val) then

; Define output file for simulation results and write header

sprintf(fname "<name_of_your_output_file")
fp_params = outfile( fname "w")
fprintf(fp_params "Corner number,Process,vdda (mV),Temperature (C),") ; header for output file
; Place  output expressions here to add to file
close(fp_params)
else

fp_params = outfile( fname "a")

; Place same output expressions here to add to file
close(fp_params)

)

sub_sim_case = sub_sim_case + 1) ; close sub_sim_case loop

sim_case = sim_case + sim_case_delta) ; close sim_case loop

exit()

Dear Yass,

My figure 1 follows:

1614.simplified_annotated_mc_params_sml_excel_122622.xlsx

Hello Shawn,

Thanks for all the help, with the code and explanation your provided I'm confident in achieving the results I want. Thanks again!

Yasso

Dear Yasso,

Wonderful! Good luck - I am sure you will have success in no time at all!

Shawn

Dear Yasso,

Wonderful! Good luck - I am sure you will have success in no time at all!

Shawn