Deduplicate strings and provide std::string_view #51098
Open
Conversation
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
…#51084) Summary: # Changelog: [Internal] Reviewed By: dannysu Differential Revision: D73850528
Summary: # Changelog: [Internal] Depending on the sampling rate, there could be lots of samples recorded. Right now, React Native requests the rate to be 10kHz. In reality, Hermes will be slower. How exactly slower will depend on the platform, because sampling profiler implementation is platform-specific. Windows peaks at ~500Hz, whereas iOS and Android that are using signals can reach up to 5kHz, based on my observations. This means that we could record sample approximately every 0.5ms. For 30 seconds of sampling, we can get around 150k samples, every one of which may have up to 500 call frames (it is a custom limitation on Hermes side). Previously, every call frame could have a unique copy of function name string and url string. This diff changes the approach to deduplicate strings, keeping a single copy for every unique string. A new abstraction `UniqueStringEntry` is added that will keep a shared pointer to the string storage and an offset to the entry in this storage. On React Native side, we are going to re-use these `std::string_view` and keep Hermes' `Profile` alive as long as they are used. We are going to achieve this by keeping unique pointer to the Hermes' `Profile` in React Native's Profile abstraction - `RuntimeSamplingProfile`, which is agnostic to the runtime. Differential Revision: D73106068
|
This pull request was exported from Phabricator. Differential Revision: D73106068 |
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Outdated suggestions cannot be applied.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.
Summary:
Changelog: [Internal]
Depending on the sampling rate, there could be lots of samples recorded. Right now, React Native requests the rate to be 10kHz. In reality, Hermes will be slower. How exactly slower will depend on the platform, because sampling profiler implementation is platform-specific. Windows peaks at ~500Hz, whereas iOS and Android that are using signals can reach up to 5kHz, based on my observations. This means that we could record sample approximately every 0.5ms.
For 30 seconds of sampling, we can get around 150k samples, every one of which may have up to 500 call frames (it is a custom limitation on Hermes side). Previously, every call frame could have a unique copy of function name string and url string.
This diff changes the approach to deduplicate strings, keeping a single copy for every unique string. A new abstraction
UniqueStringEntryis added that will keep a shared pointer to the string storage and an offset to the entry in this storage.On React Native side, we are going to re-use these
std::string_viewand keep Hermes'Profilealive as long as they are used. We are going to achieve this by keeping unique pointer to the Hermes'Profilein React Native's Profile abstraction -RuntimeSamplingProfile, which is agnostic to the runtime.Differential Revision: D73106068