(Date: 15.7.2013 – 19.7.2013)
I work a bit more on the CUDA+PandaRoot stuff. It’s working now, generally, but needs some fine tuning in the details.
On Wednesday I travel to the University of Gießen. I talk to the guys there, which also use Hough Transforms. On FPGAs and on GPUs. We exchange about our current statuses and what we plan to do in the future. They plan on finding methods to reduce the amount of data before even a Hough transformation is performed. My next step is to write a multipeak finder, if possible suitable for in-parallel execution.
We also agree on better exchanging information in the future and keeping each other up to date about any new developments and plans.
After my day in Gießen I start thinking about peak finding methods. Really convinced I’m with none. But let’s see.
In the time between that I convert my CUDA+PandaRoot research into a standalone CMake project example. Without any inter-project dependencies, just plain PandaRoot. Of course I forget to rewrite some stuff and simplifying my example takes longer then expected. But just in time on Friday I finish rewriting it.
Also on Friday (and partially on Sunday) I get an old version of PandaRoot running on Juhydra. There, the Triplet Finder code seems to run through, but producing some strange output. To be investigated next week.