-
Notifications
You must be signed in to change notification settings - Fork 5.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[CustomDevice] use CommContextManager to create xccl comm #57957
Conversation
你的PR提交成功,感谢你对开源项目的贡献! |
a2dbd4e
to
fada7ff
Compare
fada7ff
to
48302b0
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
auto store = phi::distributed::CreateOrGetGlobalTCPStore(); | ||
if (!phi::distributed::CommContextManager::GetInstance().Has( | ||
std::to_string(rid))) { | ||
phi::distributed::CommContextManager::CreateXCCLCommContext( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里看下是否需要兼容新旧通信库吧,nccl里是通过FLAGS_dynamic_static_unified_comm这个Flags判断可以做新旧通信库的切换。
124-127的判断develop分支已经更新了,这个pr好像没有提示冲突,可以手动更新下develop,防止覆盖回去了。
(当前flags的判断有问题)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
custom deivce不需要兼容旧通信库了
PR types
Others
PR changes
Others
Description
use CommContextManager to create & get xccl comm