Parallel processing for the high frame rate upgrade of the LHC synchrotron radiation telescope

The Beam Synchrotron Radiation Telescope (BSRT) is routinely used for estimating the transverse beam size, pro'le and emittance in the LHC; quantities playing a crucial role in the optimisation of the luminosity levels required by the experiments. During the 2017 LHC run, the intensi'ed analog cameras used by this system to image the beam have been replaced by GigE digital cameras coupled to image intensi'ers. Preliminary tests revealed that the typically used sub-image rectangles of 128×128 pixels can be acquired at rates of up to 400 frames per second, more than 10 times faster than the previous acquisition rate. To address the increase in CPU workload for the image processing, new VME CPU cards (Intel 4 core/2.5GHz/8GB RAM) are envisaged to be installed (replacing the previous Intel Core 2 Duo/1.5GHz/1GB RAM). This paper focuses on the software changes proposed in order to take advantage of the multi-core capabilities of the new CPU for parallel computations. It will describe how beam profile calculations can be pipe-lined through a thread pool while ensuring that the CPU keeps up with the increased data rate. To conclude, an analysis of the system performance will be presented.
Proceedings of the 16th Int. Conf. on Accelerator and Large Experimental Control Systems, ICALEPCS2017, Barcelona, Spain