-
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
Added split op bf16/fp32 oneDNN kernel #33584
Conversation
Thanks for your contribution! |
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
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
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.
It would be good if you adapted these changes to the functionality of inference BF16.
This involves:
- adding
split
to the support list in the file:std::unordered_set<std::string> supported_op_types = - adding the
mkldnn_data_type
attribute for this operator.
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
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.
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
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
PR types
New features
PR changes
OPs
Describe
Added split op bf16/fp32 oneDNN kernel. This PR caused 20% speed up on DPN68 model, measured on Intel(R) Core(TM) i9-9940X CPU @ 3.30GHz with 1000 repeated iterations. For bigger speed up keys optimizations are needed. For now OneDNN kernel will be used only if AxisTensor and SectionsTensorList are not set.