Profile reader

[iparask]

Hello Chaitanya,

in the branch feature/profreader you will find a file named profreader.py. You can use it to get the individual timings per unit and overhead for the experiments we have run. But before, considering the results correct can you test it?

If you do python profreader.py -h it will print a help message on how to use it.

[csampat]

Hi Giannis, This has to run on the child.prof.log on the client side ??

[iparask]

Hello Chaitanya, this has to run with update.0.child.prof

[csampat]

Great thanks! It works. Can you please send me the timings / profiles from your studies ? I am done with the ones I had performed

[iparask]

demPbmsims.tar.gz There you go

[csampat]

Thank you! I shall update the figure in the paper!

[csampat]

Hi, I dont know what is wrong in the code but the overhead time is coming to about 40 seconds. I guess we are missing some times during the collection in the script . I am sure RP takes more that 40-50 seconds to set up

[iparask]

Hello Chaitanya, unless you point me to the plot, I am not sure what to respond.

[csampat]

RP_plots.pdf

That faint orange line is the overhead... I am trying to do the calculations manually for one file and double checking.

If they match I will change the text in the results section for the same

[iparask]

I would be nice if you also break the DEM and PBM parts on RP. If I remember correctly only the last unit is doing PBM. That will show better what is going on.

Also, remember that those bars never had the time it takes for the Pilot to start the first unit. How you create the plot? I think it is an excel file?

[csampat]

Yes I am doing it on the excel. Yes, the timings now seem to match! So I guess the communication is low. I will change the text accordingly