In job_sort_echos and job_afni_proc_multi_echo, the computation time of the _auto_addobj was proportional to nSubj2 x nRun2 x nEcho.
The quadratic (2) effect was negligible with small & modest size dataset. However, with large dataset (nSubj > 100 & Run > 1) the computation time was atrocious. This is due do bad code optimization, which is corrected with this PR.
New cputime is proportional to nSubj x nRun x nEcho. Yay !
In
job_sort_echosandjob_afni_proc_multi_echo, the computation time of the _auto_addobj was proportional to nSubj2 x nRun2 x nEcho. The quadratic (2) effect was negligible with small & modest size dataset. However, with large dataset (nSubj > 100 & Run > 1) the computation time was atrocious. This is due do bad code optimization, which is corrected with this PR. New cputime is proportional to nSubj x nRun x nEcho. Yay !