[Fizz] Pipeable Stream Perf #24291
Merged
[Fizz] Pipeable Stream Perf #24291
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
|
Comparing: ec52a56...4fd4f93 Critical size changesIncludes critical production bundles, as well as any change greater than 2%:
Significant size changesIncludes any change greater than 0.2%: Expand to show
|
sebmarkbage
reviewed
Apr 7, 2022
sebmarkbage
reviewed
Apr 7, 2022
packages/react-server-native-relay/src/__tests__/ReactFlightNativeRelay-test.internal.js
Outdated
Show resolved
Hide resolved
acdlite
reviewed
Apr 7, 2022
|
This comment was scratch for capturing some performance timings (code that does not show up in any of the commits currently in this PR). I'll leave it here for clarity's sake but it should be disregarded.
|
Yeah. I'll add it to the list for follow-ups. For now you can inspect the output HTML to see what they do. |
gnoff
force-pushed
the
fizz-stream-perf
branch
4 times, most recently
from
abe95fc
to
39c193d
Compare
… renderToPipeableStream Modified from ssr2 and https://github.com/SuperOleg39/react-ssr-perf-test
The previous implementation of pipeable streaming (Node) suffered some performance issues brought about by the high chunk counts and innefficiencies with how node streams handle this situation. In particular the use of cork/uncork was meant to alleviate this but these methods do not do anything unless the receiving Writable Stream implements _writev which many won't. This change adopts the view based buffering techniques previously implemented for the Browser execution context. The main difference is the use of backpressure provided by the writable stream which is not implementable in the other context. Another change to note is the use of standards constructs like TextEncoder and TypedArrays.
encodeInto allows us to write directly to the view buffer that will end up getting streamed instead of encoding into an intermediate buffer and then copying that data.
|
What do pw and fcq mean? |
|
I think I'll delete that comment. it was performance.now timings for If you look in the main description you will see more broad based network results from autocannon which is more real-world |
gnoff
commented
Apr 8, 2022
| @@ -437,6 +437,7 @@ const bundles = [ | |||
| 'ReactFlightNativeRelayServerIntegration', | |||
| 'JSResourceReferenceImpl', | |||
| 'ReactNativeInternalFeatureFlags', | |||
| 'util', | |||
There was a problem hiding this comment.
@sebmarkbage you commented that this should not be necessary. In practice the build fails without this so I assume ReactServerStreamConfigNode.js is getting pulled in by accident or something changed and now that is the expected file
There was a problem hiding this comment.
The React sync for React Native is currently failing because "util could not be found", I confirmed #24322 fixes it, how can we get it landed?
sebmarkbage
approved these changes
Apr 9, 2022
rickhanlonii
pushed a commit
that referenced
this pull request
Apr 13, 2022
* Add fixture for comparing baseline render perf for renderToString and renderToPipeableStream Modified from ssr2 and https://github.com/SuperOleg39/react-ssr-perf-test * Implement buffering in pipeable streams The previous implementation of pipeable streaming (Node) suffered some performance issues brought about by the high chunk counts and innefficiencies with how node streams handle this situation. In particular the use of cork/uncork was meant to alleviate this but these methods do not do anything unless the receiving Writable Stream implements _writev which many won't. This change adopts the view based buffering techniques previously implemented for the Browser execution context. The main difference is the use of backpressure provided by the writable stream which is not implementable in the other context. Another change to note is the use of standards constructs like TextEncoder and TypedArrays. * Implement encodeInto during flushCompletedQueues encodeInto allows us to write directly to the view buffer that will end up getting streamed instead of encoding into an intermediate buffer and then copying that data.
rickhanlonii
pushed a commit
that referenced
this pull request
Apr 14, 2022
* Add fixture for comparing baseline render perf for renderToString and renderToPipeableStream Modified from ssr2 and https://github.com/SuperOleg39/react-ssr-perf-test * Implement buffering in pipeable streams The previous implementation of pipeable streaming (Node) suffered some performance issues brought about by the high chunk counts and innefficiencies with how node streams handle this situation. In particular the use of cork/uncork was meant to alleviate this but these methods do not do anything unless the receiving Writable Stream implements _writev which many won't. This change adopts the view based buffering techniques previously implemented for the Browser execution context. The main difference is the use of backpressure provided by the writable stream which is not implementable in the other context. Another change to note is the use of standards constructs like TextEncoder and TypedArrays. * Implement encodeInto during flushCompletedQueues encodeInto allows us to write directly to the view buffer that will end up getting streamed instead of encoding into an intermediate buffer and then copying that data.
zhengjitf
pushed a commit
to zhengjitf/react
that referenced
this pull request
Apr 15, 2022
* Add fixture for comparing baseline render perf for renderToString and renderToPipeableStream Modified from ssr2 and https://github.com/SuperOleg39/react-ssr-perf-test * Implement buffering in pipeable streams The previous implementation of pipeable streaming (Node) suffered some performance issues brought about by the high chunk counts and innefficiencies with how node streams handle this situation. In particular the use of cork/uncork was meant to alleviate this but these methods do not do anything unless the receiving Writable Stream implements _writev which many won't. This change adopts the view based buffering techniques previously implemented for the Browser execution context. The main difference is the use of backpressure provided by the writable stream which is not implementable in the other context. Another change to note is the use of standards constructs like TextEncoder and TypedArrays. * Implement encodeInto during flushCompletedQueues encodeInto allows us to write directly to the view buffer that will end up getting streamed instead of encoding into an intermediate buffer and then copying that data.
This was referenced Nov 8, 2024
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.
drawing heavily from #24034 this PR should improve performance of renderToPipeableStream.
This PR implements a TypedArray buffer that text is encoded into while writing chunks to the stream. It also adds a fixture which is useful for comparing the baseline rendering performance of renderToPipeableStream and renderToString.
If you want to test out the current implementation simply visit the first commit.
The second commit includes an analog of the buffer which does encoding during the performWork step.
The third commit includes a change to adopt encodeInto which delivers some performance gains by avoiding extra mem copies.
Fixture Performance
The following stats were gathered with
npx autocannon -c 5 -d 30 http://localhost:4000/{stream|string} --renderStatusCodes --excludeErrorStatsBefore Change
buffering stream chunks
encodeInto