[oneDNN] Improving Graph Rewrite Performance #50932
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@penpornk Can you please review this PR ? Thanks! |
penpornk
requested changes
Sep 14, 2021
| bool kernel_registered = false; | ||
|
|
||
| if (kernel_element == registered_kernels_map.end()) { | ||
| string kernel = KernelsRegisteredForOp(op_name); |
There was a problem hiding this comment.
Nit: Please make the variable name more meaningful for readability.
Suggested change
| string kernel = KernelsRegisteredForOp(op_name); | |
| string registered_kernels = KernelsRegisteredForOp(op_name); |
Comment on lines
234
to
236
| string search_string = label; | ||
| search_string += string(";") + string(" T in ["); | ||
| search_string += DataType_Name(T) + string("]"); |
There was a problem hiding this comment.
Nit: This can be done in one command. (Please help format the source. The line is probably too long.)
Suggested change
| string search_string = label; | |
| search_string += string(";") + string(" T in ["); | |
| search_string += DataType_Name(T) + string("]"); | |
| string search_string = label + string("; T in [") + DataType_Name(T) + string("]"); |
Comment on lines
151
to
153
| // Map used to store kernel registration status | ||
| thread_local static auto* registered_kernels_map = | ||
| new absl::flat_hash_map<string, bool>(); |
There was a problem hiding this comment.
Does this need to be a global static variable? We try to only have static variable when they are trivially destructible. Can this map be part of some class? Or be inside of IsMklOp()?
There was a problem hiding this comment.
yes it can be in IsMklOp(). I made that change.
|
Thank you for reviewing the PR. I have addressed your comments. |
google-ml-butler
bot
added
kokoro:force-run
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR improves the latency of graph rewrite by avoiding doing kernel registry lookup more than once per kernel. Similar fix was to done before to eager mode, so the refactoring is done to manage and maintain that kernel registry info in one place (mkl_graph_util.h)