The SPRINT famework was ported to HECToR, the UK National Supercomputing Service.
The implementation of the pcor and pmaxT functions have been optimised on HECToR.
The previous SPRINT parallel implementation of the pair-wise Pearson’s correlation function uses a task farm approach where the master process assigns tasks to slave processes and collects the results when the tasks are finished as described above. This code had been analysed and optimised to enable it to scale to 512 slave processes and beyond. This new code has been succesfully tested and benchmarked on the UK National Supercomputing Service, the HECToR Cray XT system.
The SPRINT parallel implementation of the permutation testing function mt.maxt scales almost perfectly on HECToR. Benchmarks on other systems show sufficient speed-up that reasonable performance gains can be had even in environments where network latency can impact.
SPRINT: Parallel computing with R on HECToR - Training Course
The first ever SPRINT training course took place at NAG Oxford on December 1st 2011. The course aimed to give a general overview of the hardware, software and configuration of the HECToR system in the context of using SPRINT.
This work was funded by an HECToR computational science and engneering support award (dCSE) administered by NAG Ltd