Store instruction annotation values as C++ doubles.#366
Open
virajbshah wants to merge 3 commits intogoogle:mainfrom
Open
Store instruction annotation values as C++ doubles.#366virajbshah wants to merge 3 commits intogoogle:mainfrom
doubles.#366virajbshah wants to merge 3 commits intogoogle:mainfrom
Conversation
* Factor out the per-instruction add to graph logic. * This change is in preparation for further changes to the graph construction logic. * Specifically, this should make the trace graph construction logic cleaner.
* Move annotation adding logic to `AddInstructionAnnotations`. * Remove default annotation magic number.
* Instruction annotations are currently stored as `float`s in the C++ parts of the codebase. * This matches the type currently used by the models (`tf.dtypes.float32`), but not the proto storage type (`double`). * For consistency, store annotations as `double`s at the C++ end, then cast them to whatever type the models require.
Collaborator
boomanaiden154
left a comment
boomanaiden154
left a comment
There was a problem hiding this comment.
What's the benefit of doing the narrowing to float later versus earlier? It doesn't seem like this patch really enables anything and it will increase memory consumption.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
floats in the C++ graph builder.tf.dtypes.float32), but not the proto storage type (double).doubles at the C++ end, then narrow down to the type the models use.