R is usually said to be the single-threaded software by default. Therefore, one sometimes needs too much time to run some heavy codes. However, there are some packages available in R which allow to run on a multiple thread on the CPU. My question is how to utilize multiple cores in a machine? I would be appreciated if anyone can share his or her experience with the multi-core computations using R.
I think that R in and of itself is not multi-threaded, but like you stated, individual packages may use multi-threading/multi-processing for some operations. I guess the question would be what problems are you trying to solve, and would parallel algorithms work better on those problems? Some problems just don't lend themselves well to parallel computation.
A number of people are working on different R implementations to exploit multi-core systems, but they may or may not be useful depending on what you're working on.
Sorry to be terribly non-specific, but it all kind of depends on the problem you're trying to solve.
If you consider the throughput, then multicore can run multiple instances simultaneously and harnessing the computing power of the available cores.
If you consider the latency, then the entire R computing should be re-designed and re-implemented (i.e., be multi-threaded program).
- Badges
- Science topic
- Similar topics
- Automotive Engineering
- Machines