I was wondering if it might be possible to modify samr() to be computed in a parallelized manner?
With the current single processor code, running on a large-ish (~50,000 x 750) dataset takes quite a long time, even on a modern machine (going on 3 days now for a survival model via SAMseq()..)
At a glance (I haven't done any profiling..), it seems like there are a few for-loops in the code for samr() that could be parallelized using something like foreach.
Greetings!
I was wondering if it might be possible to modify
samr()
to be computed in a parallelized manner?With the current single processor code, running on a large-ish (~50,000 x 750) dataset takes quite a long time, even on a modern machine (going on 3 days now for a survival model via
SAMseq()
..)At a glance (I haven't done any profiling..), it seems like there are a few for-loops in the code for
samr()
that could be parallelized using something like foreach.Is this something that has been considered?