[GitHub] [incubator-tvm] yidawang commented on issue #4344: [ThreadPool] Solve thread transitions issue

[GitBox <gi...@apache.org>]

yidawang commented on issue #4344: [ThreadPool] Solve thread transitions issue
URL: https://github.com/apache/incubator-tvm/pull/4344#issuecomment-555386180
 
 
   > Hmm...I think Auto Tuning and OpenCV + TVM is very common and many users will forget this things, so I prefer it be `false` by default. However, we could add comment on `exclude_worker_0` and suggest to be `true` when to use single thread. Also open to listen more people's comment. @vinx13 @tqchen guys could tag anyone interested.
   
   At least the Auto tuning case is about binding the master thread or not, which we have discussed [here](https://discuss.tvm.ai/t/use-tvm-darknet-to-detect-vidoes-after-relay-build-module-build-cv2-ops-cost-much-more-time/4730/25). After reviewing all these use cases, I think by default we probably 1) don't bind the master thread; 2) exclude work0, i.e. bind the first task to core 0.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services