First steps to CINECA – Solving the mystery of ParaView Catalyst
It was a wonderful experience in SoHPC 2016 first week training in Fz-Juelich super computing center where I attended lectures on MPI, OpenMP, CUDA, and ParaView visualizations. And I added some new friends in my life. It was really interesting for me to meet new people from different parts of Europe under one roof ,thanks to Summer Of HPC. We enjoyed a lot together. After Friday night fare well party it was time to say ‘sayonara’ means to say goodbye to each other and to pack our bags to fly to our respective project sites.With full of excitement and enthusiasm, I landed in Bologna (my project site) beautiful lively city, with nice people and with one super computing center named – Cineca.
Cineca, my work place having four supercomputers Pico, Galileo, Marconi and Fermi. And me and my colleague Petr got the opportunity to start our work with Pico. We started our work from reading a so called ParaView Catalyst user guide luckily it was only 66 pages book, and not so boring. While reading I got engrossed in it and I finished reading it in two days but did not understand every part of it, then luckily on Tuesday we met Paolo Lazzari who showed us some python codes for ParaView which generates some visualizations and he gave us some tasks to play with that code.Then on Wednesday we met Silvano , VTK expert and he discussed about VTK libraries and basic functions which we have to use in our project.But still I was in dilemma and thinking how we are going to run examples using ParaView Catalyst. So it was the time to solve the mystery of ParaView Catalyst.
On Thursday Petr and I decided to run some ParaView Catalyst examples. So I decided that I have to run the examples on Pico , the initial problem I saw was the old ParaView version, so I downloaded the new ParaView Source version on Pico and trying to install it but there was some problem with qt version while building it.I tried a lot sorting it, installed older version of qt. But still the same errors. Time was passing by and two days got over. Then after thoughtful discussion with my colleague Petr, I decided to build ParaView without qt. It worked fine. Then I started working with the examples. After loading proper modules I was able to compile and run my examples. But I was more excited to see the Visualization output on ParaView GUI, for that I have to fix the issue of connecting two different nodes – one for simulation and other for visualization. It was as usual ‘The generation gap’ problem, between the nodes so I downloaded the new ParaView binary installer version and brought them to the same age group. In the end they got connected, but still no visualization. I was trying every possibility which was coming in my mind. It was very important for us to run a catalyst example on Pico. I discussed the problems with Luigi Calori, Silvano and Massimiliano. The problem was the simulation code was running fine on one node but it was not able to communicate with the visualization node. On Thursday Massimiliano also showed us the simulation code which I have to use in my project. One week 4 days got over , still I was not able to see the running visualization, while going home I was sad and I said to myself I have one more day to fix this. I took a deep breathe and just relax.
Friday morning with fresh brain I tried again. I read again some documentations on pvserver and I got a new idea for Pico and discussed with Luigi Calori. We tried it and after doing some changes in some files. It worked, I was so happy at that moment, got motivation to work further. After lunch we met Francesca Delli Ponti, to learn some basics of Blender a new software to render images and to make videos from visualizations. I liked it.
So this was my story of solving the mystery of ParaView Catalyst and I must say “Be patient, be stubborn and stick to problem. Definitely you are gonna solve it”.
My first blender made video “Introduction to Cineca and my colleagues”.Check this out.