Is it possible to read data from unfinished simulation?
I have run a simulation which took me about 6 hourse but I terminated it before it was completed. The problem is I defined a nodal output frecuency higher than the number of iterations run, so when I load the log file it says that data do not exist as the data of the last iteration was not stored.
is it possible to retrieve nodal data even if the nodal output frecuency (1000) was not reached?
Best Answer
-
The requested output files are written per the stated/requested frequency - or when the job does a clean stop. If you killed the job, rather than requesting it to stop with a clean finish - no result files would be written.
When running from one of the GUIs you should have a 'Stop' button - this will stop the job at the end of the time step when it receives the signal. (Command-line acuSig -stop will do the same thing.)
There's probably also an 'Output' button - which will write requested output at the end of the time step when it receives that signal. (Command-line acuSig -out will do the same thing.)
You should still have access to the 'default' surface integrated data for 'Plot' tool or AcuProbe - just not the nodal output for visualization.
0
Altair EmployeeAnswers
-
The requested output files are written per the stated/requested frequency - or when the job does a clean stop. If you killed the job, rather than requesting it to stop with a clean finish - no result files would be written.
When running from one of the GUIs you should have a 'Stop' button - this will stop the job at the end of the time step when it receives the signal. (Command-line acuSig -stop will do the same thing.)
There's probably also an 'Output' button - which will write requested output at the end of the time step when it receives that signal. (Command-line acuSig -out will do the same thing.)
You should still have access to the 'default' surface integrated data for 'Plot' tool or AcuProbe - just not the nodal output for visualization.
0 -
acupro_21778 said:
The requested output files are written per the stated/requested frequency - or when the job does a clean stop. If you killed the job, rather than requesting it to stop with a clean finish - no result files would be written.
When running from one of the GUIs you should have a 'Stop' button - this will stop the job at the end of the time step when it receives the signal. (Command-line acuSig -stop will do the same thing.)
There's probably also an 'Output' button - which will write requested output at the end of the time step when it receives that signal. (Command-line acuSig -out will do the same thing.)
You should still have access to the 'default' surface integrated data for 'Plot' tool or AcuProbe - just not the nodal output for visualization.
Thanks acupro. I just killed the job, so no files were written. Thanks for the tips about the stop button, I will take it into account next time.
0 -
acupro_21778 said:
The requested output files are written per the stated/requested frequency - or when the job does a clean stop. If you killed the job, rather than requesting it to stop with a clean finish - no result files would be written.
When running from one of the GUIs you should have a 'Stop' button - this will stop the job at the end of the time step when it receives the signal. (Command-line acuSig -stop will do the same thing.)
There's probably also an 'Output' button - which will write requested output at the end of the time step when it receives that signal. (Command-line acuSig -out will do the same thing.)
You should still have access to the 'default' surface integrated data for 'Plot' tool or AcuProbe - just not the nodal output for visualization.
I pressed stop from the manage icon in gui interface but the simulation does not stop--it continues to run till the full time step and then only stops. I specified the simulation for 100 time steps. I tried writing the output but it does not write in the middle of simulation--it writes only after the simulation is over. I checked the log file--it ran to full simulation of 100 time step.
Am i doing anything wrong?
0 -
Prabin Pradhananga_20428 said:
I pressed stop from the manage icon in gui interface but the simulation does not stop--it continues to run till the full time step and then only stops. I specified the simulation for 100 time steps. I tried writing the output but it does not write in the middle of simulation--it writes only after the simulation is over. I checked the log file--it ran to full simulation of 100 time step.
Am i doing anything wrong?
Which GUI are you using? Are you sure you're pressing stop/output for the correct run id? Can you please attach an image showing how you press the stop/output button?
Did you also try from the Command Prompt - with auSig -stop or acuSig -out issued from the correct directory? (Add problem name, runID, etc as needed.)
0 -
acupro_21778 said:
Which GUI are you using? Are you sure you're pressing stop/output for the correct run id? Can you please attach an image showing how you press the stop/output button?
Did you also try from the Command Prompt - with auSig -stop or acuSig -out issued from the correct directory? (Add problem name, runID, etc as needed.)
Yes I used both gui and command line for the stop but did not work. I simply right click and press stop. Does not work.
0 -
Prabin Pradhananga_20428 said:
Yes I used both gui and command line for the stop but did not work. I simply right click and press stop. Does not work.
Can you attach the Log file for that Run 32? Which version of HyperWorks CFD are you using? And on which OS - Linux or Windows?
0 -
Prabin Pradhananga_20428 said:
Yes I used both gui and command line for the stop but did not work. I simply right click and press stop. Does not work.
Can you attach the Log file for that Run 32? Which version of HyperWorks CFD are you using? And on which OS - Linux or Windows?
0 -
acupro_21778 said:
Can you attach the Log file for that Run 32? Which version of HyperWorks CFD are you using? And on which OS - Linux or Windows?
I am using version 2022.1 in linux machine.
0 -
Prabin Pradhananga_20428 said:
I am using version 2022.1 in linux machine.
Based on your Log file, I can say that for "some" reason, acuSig command (or "stop" if using GUI) was not correctly executed. The exact reason is difficult to tell just based on Log file.
If "Stop" (or acuSig -stop) or "write output" (or acuSig -out), were correctly executred
then the Log file would have "user signaled to output" like in the screenshot below.
1 -
ydigit_21275 said:
Based on your Log file, I can say that for "some" reason, acuSig command (or "stop" if using GUI) was not correctly executed. The exact reason is difficult to tell just based on Log file.
If "Stop" (or acuSig -stop) or "write output" (or acuSig -out), were correctly executred
then the Log file would have "user signaled to output" like in the screenshot below.
Thanks for sharing!
0 -
Prabin Pradhananga_20428 said:
I am using version 2022.1 in linux machine.
It looks like you may be specifying a different problem directory than where you have saved the HyperWorks CFD file, since there is a '-pdir' flag in the acuRun command. This may 'break' some of the automatic processes.
If you save the HyperWorks CFD file to particular directory, then make sure that same directory is the default location for project directory - then run - does the stop or output button then work as expected?
0 -
acupro_21778 said:
It looks like you may be specifying a different problem directory than where you have saved the HyperWorks CFD file, since there is a '-pdir' flag in the acuRun command. This may 'break' some of the automatic processes.
If you save the HyperWorks CFD file to particular directory, then make sure that same directory is the default location for project directory - then run - does the stop or output button then work as expected?
It is not working. I did it multiple times using GUI and acuSig -pdir /home/prp20007/periodic_tiles -stop but it never stops.
0 -
Prabin Pradhananga_20428 said:
It is not working. I did it multiple times using GUI and acuSig -pdir /home/prp20007/periodic_tiles -stop but it never stops.
Could you attach an image with your directory structure?
Do you see test.inp (and the several .Log files) in the /home/prp20007/test/ directory?
Do you also see ACUSIM.DIR directory in the /home/prp20007/test/ directory?Where is the .hm file saved - in which directory?
0 -
acupro_21778 said:
Could you attach an image with your directory structure?
Do you see test.inp (and the several .Log files) in the /home/prp20007/test/ directory?
Do you also see ACUSIM.DIR directory in the /home/prp20007/test/ directory?Where is the .hm file saved - in which directory?
Yes, the log files, input files, .hm files and ACUSIM dir are all in the same folder.
0 -
Prabin Pradhananga_20428 said:
Yes, the log files, input files, .hm files and ACUSIM dir are all in the same folder.
Can you also attach the Acusim.cnf file - or post the content?
When you tried the acuSig commands from command-line, were you already in that same .../test directory?
0 -
acupro_21778 said:
Can you also attach the Acusim.cnf file - or post the content?
When you tried the acuSig commands from command-line, were you already in that same .../test directory?
I opened a different terminal in the same working directory and used acuSig -stop. It does not stop.
0 -
Prabin Pradhananga_20428 said:
I opened a different terminal in the same working directory and used acuSig -stop. It does not stop.
If you issue:
ls ACUSIM.DIR/*sig
you should see a .sig file for each runCan you please share the content of the .sig file for the run where you just issued the command-line acuSig -stop
The content might be something like this:
more ACUSIM.DIR/*9.sig
acuSolve_control = output
acuSolve_control = stop0 -
acupro_21778 said:
If you issue:
ls ACUSIM.DIR/*sig
you should see a .sig file for each runCan you please share the content of the .sig file for the run where you just issued the command-line acuSig -stop
The content might be something like this:
more ACUSIM.DIR/*9.sig
acuSolve_control = output
acuSolve_control = stop
I could see stop signal here. I checked log file. No sign of 'user signaled to stop'.
0 -
Prabin Pradhananga_20428 said:
I could see stop signal here. I checked log file. No sign of 'user signaled to stop'.
Yes - it does look like the process is working correctly to send the signal - it's just not being received on the solver side.
I looked back at the 32.Log file - and saw the job is running very quickly. Perhaps that is the reason, there's just not the time for the signal send/receive process to occur.
I also see the mesh is very small - only 320000 nodes. We recommend on the order of minimum 10000 to 15000 mesh nodes per core. I would expect good/linear scaling in the 20-32 core range. Beyond that, the message passing will start to hinder the efficiency of the simulation. There might still be speedup, but it would likely not be linear.
Could you try this same model with 2 cores (get it run more slowly), and see if there is now time for the stop or output signal to get 'caught' by the solver processes?
1 -
acupro_21778 said:
Yes - it does look like the process is working correctly to send the signal - it's just not being received on the solver side.
I looked back at the 32.Log file - and saw the job is running very quickly. Perhaps that is the reason, there's just not the time for the signal send/receive process to occur.
I also see the mesh is very small - only 320000 nodes. We recommend on the order of minimum 10000 to 15000 mesh nodes per core. I would expect good/linear scaling in the 20-32 core range. Beyond that, the message passing will start to hinder the efficiency of the simulation. There might still be speedup, but it would likely not be linear.
Could you try this same model with 2 cores (get it run more slowly), and see if there is now time for the stop or output signal to get 'caught' by the solver processes?
It does not work either.
0