Using the Parallel Collider
Hi Yade programmers and users,
I have been trying to to shorten the run time of my code on a multi-CPU cluster. As I was going through past Q&A's on the forum, I noticed discussions about the parallel collider that was tested a while ago in [1] . However, I wasn't able to find out clearly the state of the parallel collider after those tests in 2012.
My questions are:
1. Has the parallel collider already been pushed into the trunk version? Or is there a separate version of Yade that I need to use the collider?
2. Are there any specific commands to call the collider or would yade -j suffice?
Thank you for helping me in advance!!
Best,
Yuxuan
Question information
- Language:
- English Edit question
- Status:
- Solved
- For:
- Yade Edit question
- Assignee:
- No assignee Edit question
- Solved by:
- Bruno Chareyre
- Solved:
- 2021-03-24
- Last query:
- 2021-03-24
- Last reply:
- 2021-03-23
Hi,
1. it is in trunk
2. nothing specific
If you want to see how it works you can do this:
O.timing.
O.run(N,True)
from yade import timing
timing.stats()
You will see how much is spent on each engine. Then if you run with different -j you will see how the different engines react to that.
Usually collider is only a few percent of the total.
Regards
Bruno
Yuxuan Wang (ywang1) said : | #2 |
Thanks Bruno! That solved my question.