exit when allreduce/broadcast error cause timeout (#112)

* keep async timeout task

* add missing pthread to cmake

* add tests

* Add a sleep period to avoid flushing the tracker.
This commit is contained in:
Chen Qin
2019-10-11 00:39:39 -07:00
committed by Jiaming Yuan
parent af7281afe3
commit 5d1b613910
17 changed files with 403 additions and 71 deletions

3
.gitignore vendored
View File

@@ -47,3 +47,6 @@ mpich-3.2/
cmake-build-debug/
.vscode/
# cmake
build/